Form DOT F (8-72) Technical Report Documentation Page. 2. Government Accession No. 3. Recipient's Catalog No. 1. Report No. FHWA/TX-11/

Transcription

1 1. Report No. FHWA/TX-11/ Government Accession No. 3. Recipient's Catalog No. 4. Title and Subtitle DEVELOPMENT OF GUIDELINES FOR PEDESTRIAN SAFETY TREATMENTS AT SIGNALIZED INTERSECTIONS Technical Report Documentation Page 5. Report Date August 2011 Published: January Performing Organization Code 7. Author(s) James Bonneson, Michael Pratt, and Praprut Songchitruksa 9. Performing Organization Name and Address Texas Transportation Institute The Texas A&M University System College Station, Texas Sponsoring Agency Name and Address Texas Department of Transportation Research and Technology Implementation Office P.O. Box 5080 Austin, Texas Performing Organization Report No. Report Work Unit No. (TRAIS) 11. Contract or Grant No. Project Type of Report and Period Covered Technical Report: September 2009-August Sponsoring Agency Code 15. Supplementary Notes Project performed in cooperation with the Texas Department of Transportation and the Federal Highway Administration. Project Title: Development of Pedestrian Safety-Based Warrants for Protected or Protected-Permissive Left-Turn Control URL: Abstract For intersections with a permissive or protected-permissive left-turn mode, pedestrians cross during the permissive period. This operation requires the left-turn driver to yield to both opposing vehicles and pedestrians, prior to accepting a gap and completing the turn. Pedestrian crash risks are increased in these complicated driving conditions because left-turn drivers sometimes fail to yield to pedestrians. This document summarizes the research conducted and the conclusions reached during the development of guidelines for pedestrian safety treatments at signalized intersections. The guidelines are focused on treatments that alleviate conflicts between left-turning vehicles and pedestrians. One treatment addressed in the document is the use of protected or protected-permissive left-turn operation. The guidelines are based on consideration of pedestrian safety and vehicle operation. These considerations include the road-user costs associated with pedestrian-vehicle crashes and vehicle delay. The guidelines were incorporated in the Traffic Signal Operations Handbook. The guidelines were also incorporated into a spreadsheet that was developed to accompany the Handbook. The Handbook was previously developed for Project It provides guidelines for timing traffic control signals at intersections that operate in isolation or as part of a coordinated signal system. 17. Key Words Signalized Intersections, Intersection Design, Intersection Performance, Traffic Signal Timing, Pedestrian Safety 19. Security Classif.(of this report) Unclassified Form DOT F (8-72) 20. Security Classif.(of this page) Unclassified Reproduction of completed page authorized 18. Distribution Statement No restrictions. This document is available to the public through NTIS: National Technical Information Service Alexandria, Virginia No. of Pages Price

2

3 DEVELOPMENT OF GUIDELINES FOR PEDESTRIAN SAFETY TREATMENTS AT SIGNALIZED INTERSECTIONS by James Bonneson, P.E. Senior Research Engineer Texas Transportation Institute Michael Pratt, P.E. Assistant Research Engineer Texas Transportation Institute and Praprut Songchitruksa, P.E. Associate Research Engineer Texas Transportation Institute Report Project Project Title: Development of Pedestrian Safety-Based Warrants for Protected or Protected-Permissive Left-Turn Control Performed in cooperation with the Texas Department of Transportation and the Federal Highway Administration August 2011 Published: January 2012 TEXAS TRANSPORTATION INSTITUTE The Texas A&M University System College Station, Texas

4

5 DISCLAIMER The contents of this report reflect the views of the authors, who are responsible for the facts and the accuracy of the data published herein. The contents do not necessarily reflect the official view or policies of the Federal Highway Administration (FHWA) and/or the Texas Department of Transportation (TxDOT). This report does not constitute a standard, specification, or regulation. It is not intended for construction, bidding, or permit purposes. The engineer in charge of the project was James Bonneson, P.E. # NOTICE The United States Government and the State of Texas do not endorse products or manufacturers. Trade or manufacturers names appear herein solely because they are considered essential to the object of this report. v

6 ACKNOWLEDGMENTS This research project was sponsored by the Texas Department of Transportation and the Federal Highway Administration. The research was conducted by Dr. James Bonneson, Mr. Michael Pratt, and Dr. Praprut Songchitruksa. All three researchers are with the Texas Transportation Institute. The researchers acknowledge the support and guidance provided by the Project Monitoring Committee:! Mr. Adam Chodkiewicz, Project Director (TxDOT, Traffic Operations Division).! Mr. James Bailey (TxDOT, Waco District).! Mr. Scott Cunningham (TxDOT, Austin District).! Mr. John Gianotti (TxDOT, San Antonio District).! Mr. Jianming Ma (TxDOT, Traffic Operations Division).! Ms. Wendy Simmons (TxDOT, Tyler District).! Mr. Wade Odell, Research Engineer (TxDOT, Research and Technology Implementation Office). The researchers also acknowledge the contribution of Dr. Kay Fitzpatrick to this project. Dr. Fitzpatrick provided direction in this document s development and reviewed several of its early draft versions. The researchers are grateful to the practitioners that participated in the state-of-the-practice interviews, on the expert panel, and in the pilot workshop that demonstrated the proposed guidelines. They are particularly grateful to Mr. Ali Mozdbar, Traffic Signal System Manager, with the Transportation Department of the City of Austin. He implemented pedestrian safety treatments at several intersections in Austin for the purpose of conducting a before-after observational study of treatment effectiveness. vi

7 TABLE OF CONTENTS Page LIST OF FIGURES... viii LIST OF TABLES... ix CHAPTER 1. INTRODUCTION OVERVIEW OBJECTIVES RESEARCH APPROACH TERMINOLOGY REFERENCES CHAPTER 2. LITERATURE REVIEW OVERVIEW PEDESTRIAN SAFETY IN TEXAS INTERSECTION TRAFFIC CONTROL LEFT-TURN-RELATED SAFETY ISSUES CANDIDATE TRAFFIC ENGINEERING TREATMENTS REFERENCES CHAPTER 3. STATE OF THE PRACTICE REVIEW OVERVIEW PRACTITIONER INTERVIEW EXPERT PANEL DISCUSSION CHAPTER 4. MODEL DEVELOPMENT AND DATA COLLECTION OVERVIEW OPERATIONS MODEL AND DATA COLLECTION PLAN SAFETY MODEL AND DATA COLLECTION PLAN REFERENCES CHAPTER 5. DATA REDUCTION AND SUMMARY OVERVIEW OPERATIONS DATA SAFETY DATA CHAPTER 6. MODEL CALIBRATION AND GUIDELINE DEVELOPMENT OVERVIEW SAFETY PREDICTION MODEL CALIBRATION ROAD-USER COST PREDICTION MODEL DEVELOPMENT GUIDELINE DEVELOPMENT REFERENCES APPENDIX A. STUDY SITE CHARACTERISTICS... A-1 APPENDIX B. BEFORE-AFTER EVALUATION OF SELECTED TREATMENTS..B-1 vii

8 LIST OF FIGURES Figure Page 1-1 Intersection Conflict Areas Guidelines for Determining Left-Turn Mode Pedestrian Level of Service Based on Traffic Volume Crash Frequency Based on Traffic Volume Crash and Conflict Frequency for Permissive Operation Crash and Conflict Relationship for Permissive Operation Crash and Conflict Frequency for Unopposed Left-Turn Operation Crash and Conflict Relationship for Unopposed Left-Turn Operation Relationship between Pedestrian Volume and Vehicle Delay Interview Questions for Texas Agencies Interview Questions for Agencies Outside of Texas E-Discussion Questions Factors Contributing to Conflicts Rank of Selected Treatments Video Camera Locations Comparison of Observed and Predicted Legal Conflict Frequency Comparison of Observed and Predicted Illegal Conflict Frequency Comparison of Observed and Predicted Legal Pedestrian Volume Left-Turn Mode Selection Based on Pedestrian Safety viii

9 LIST OF TABLES Table Page 2-1 Pedestrian Crash Statistics Pedestrian Clearance Time Traffic Engineering Treatments for Improving Pedestrian Safety at Intersections Public Agencies Represented in the Practitioner Interview Crosswalk Provision Trends at Urban Signalized Intersections Classification of High-Pedestrian-Volume Intersections Public Agencies Represented in the Expert Panel Discussion Conflict Occurrence Response Analysis Comparison of Simulation Tool Capabilities Test Bed Geometric and Signalization Characteristics Test Bed Traffic Characteristics Test Bed Modifications for Each Pedestrian Treatment Variables in Conflict Database Study Site Description Summary of Intersection Delay Statistics by Treatment Percent Change in Intersection Delay by Treatment Cycle and Pedestrian Counts by Site Category Legal Pedestrian Distribution by Gender and Age Group Vehicular Volumes by Site Category Pedestrian Volumes by Site Category Pedestrian Volumes and Platoon Patterns Pedestrian-Based Conflicts by Site Category Calibrated Model Statistical Description Legal Pedestrian-Vehicle Conflict Model Calibrated Model Statistical Description Illegal Pedestrian-Vehicle Conflict Model Calibrated Model Statistical Description Legal Pedestrian Volume Model Crash Distribution at Urban Signalized Intersections in Texas Study Site Left-Turn-Related Pedestrian-Vehicle Safety Statistics Left-Turn-Related Pedestrian-Vehicle Conflict Analysis Left-Turn-Related Pedestrian-Vehicle Crash Analysis Average Cost of a Left-Turn-Related Pedestrian-Vehicle Crash at an Urban Signalized Intersection Average Value of Travel Time Effect of Pedestrian Treatments on Delay and Conflicts at Intersections ix

10

11 CHAPTER 1. INTRODUCTION OVERVIEW There are about 6000 pedestrian-involved crashes each year in Texas. This number is small compared to the 300,000 vehicle-vehicle crashes in Texas. However, an examination of pedestrianrelated crash rates on a nationwide basis indicates that Texas ranks in the top 16 states. This trend suggests that there is potential for pedestrian safety improvement on Texas highways. Closer examination of the data indicate that this trend extends to pedestrian safety at signalized intersections in Texas. For intersections with a permissive or a protected-permissive left-turn mode, pedestrians cross during the permissive period along with the parallel through vehicular movement. This requires the left-turn driver to yield to both opposing vehicles and pedestrians prior to accepting an appropriate gap. Pedestrian crash risks are increased in these complicated driving conditions because left-turn drivers sometimes make misjudgments and fail to yield to pedestrians. Existing left-turn mode selection guidelines focus mainly on the vehicular traffic conditions at the intersection. Few of them include specific consideration of pedestrian safety. For example, existing guidelines for protected-permissive control typically focus on the left-turn and opposing through traffic volumes. Very few of these guidelines include a sensitivity to pedestrian volume or other pedestrian safety-related factors (e.g., crossing distance, median presence and width, driver sight distance, pedestrian compliance). OBJECTIVES The primary objective of this project was to incorporate pedestrian safety considerations into the guidelines for selecting left-turn operational mode (i.e., protected, protected-permissive, or permissive). A secondary objective was to develop guidelines addressing a broader range of treatments that may be used to improve pedestrian safety at signalized intersections. The focus of the research was on conflicts between left-turning vehicles and pedestrians. To achieve these objectives, the research was comprehensive in its consideration of pedestrian safety issues at signalized intersections in Texas. The guidelines were documented in two forms. Guidelines to assist in the determination of the appropriate left-turn mode were documented in the report titled Pedestrian Safety Guidelines and Proposed Left-Turn Phase Warrant. This document serves as a quick reference guide that identifies conditions where protected or protectedpermissive operation is a cost-effective treatment for pedestrian-related safety problems. Guidelines were also documented in an updated version of the Traffic Signal Operations Handbook (previously developed for TxDOT in Project ). The guidelines that were added to the Handbook address changes to the signal timing, phase sequence, or pedestrian crossing distance to minimize left-turn-related pedestrian-vehicle conflicts and improve pedestrian safety. 1-1

12 RESEARCH APPROACH A two-year program of research was developed to satisfy the project s research objectives. During the first year of research, city, county, and state engineers were interviewed, and field data were collected to ascertain the range of pedestrian safety concerns at intersections. The data were subsequently used to quantify the impacts of alternative left-turn modes (and other treatments) on safety and efficiency. During the second year, the field data and input from an expert panel were used to develop guidelines that are sensitive to a wide range of intersection conditions, and are applicable on a statewide basis. The research approach consists of eight tasks that represent a logical sequence of review, research, evaluation, and workshop development. These tasks are identified in the following list. 1. Finalize Work Plan. 2. Evaluate State-of-the-Practice. 3. Develop Data Collection Plan and Collect Data. 4. Reduce Data. 5. Develop and Evaluate Preliminary Guidelines. 6. Conduct Before-After Study. 7. Conduct Pilot Workshop and Revise Guidelines. 8. Prepare Research Report. The research conducted in Tasks 1 through 8 is documented in this report. TERMINOLOGY This section defines various terms that are used in this report. Crosswalk Area The crosswalk area is the area of pavement outlined by the curb line and a line offset outwardly from (and parallel to) the crosswalk by 10 ft. Thus, the crosswalk area for a crosswalk that is 12 ft wide is effectively 32 ft wide (= ). This definition recognizes that some pedestrians crossing at the intersection do not walk fully within the marked crosswalk for some or all of their crossing, but are considered to be compliant with the crosswalk s intended purpose (1). Pedestrian-Vehicle Conflict A pedestrian-vehicle conflict is defined as an event where the projected path of a vehicle and a pedestrian cross and either the pedestrian or the vehicle, or both, make a last-second change in their path direction, speed, or both to avoid a collision. This definition is consistent with that used by several researchers (2, 3). Lord (4) compared crash data and conflict data for common intersections, where conflicts were defined using this definition. He found very good agreement (R 2 = 0.59) between the two statistics. 1-2

13 In their definition of conflict, Carter et al. (5) considered the time span associated with the change in path direction, speed, or both. They consider the most sudden changes in course to be a conflict and the non-sudden (i.e., slower) changes to be an avoidance maneuver. However, during their field studies, they found it difficult to distinguish between these two categories and chose to combine them for purposes of pedestrian safety assessment. For this reason, the aforementioned definition of pedestrian-vehicle conflict is defined to include avoidance maneuvers. Pedestrian-Based Conflict A pedestrian-based conflict is a conflict where a pedestrian in the crosswalk changes his or her path direction, speed, or both to avoid a vehicle. The vehicle may, or may not, alter its course. For a given left-turn vehicle, there can be several conflicts if several pedestrians alter their course. In the extreme, the number of pedestrian-based conflicts is equal to the product of the pedestrian volume and the vehicular volume (i.e., every pedestrian is conflicted by every vehicle). Typical pedestrian-based conflicts include (5):! Pedestrian stepped into roadway and then stepped back onto the curb to let vehicle pass.! Pedestrian went around vehicle that was blocking crosswalk.! Pedestrian hurried while crossing to avoid oncoming vehicle.! Pedestrian stopped (or noticeably slowed) while crossing to let vehicle cross.! Pedestrian delayed leaving the origin curb due to a vehicle.! Pedestrian attempted to cross but did not leave the origin curb. Vehicle-Based Conflict A vehicle-based conflict is a conflict where one or more vehicles change path direction, speed, or both to avoid pedestrians and no pedestrians alter their course. A second vehicle may be involved in this conflict (e.g., same lane, opposing through). In the extreme, the number of vehicle-based conflicts is equal to the subject vehicle volume. Pedestrian Signal Violation A pedestrian signal violation occurs when a pedestrian begins the crossing (i.e., steps from behind the curb into the crosswalk) during the flashing DON T WALK or the steady DON T WALK indications. Following field measurement of pedestrian signal violations, Kattan et al. (6) observed that some pedestrians could enter the crosswalk 2 to 3 s after the end of the WALK indication and still complete the crossing before the end of the flashing DON T WALK indication. They rationalized that, while pedestrians in this group were technically violating the signal, they were not associated with as high a safety risk as those pedestrians that started crossing well after the end of the WALK indication. Pedestrian-Vehicle Conflict Area When evaluating conflicts between left-turning vehicles and pedestrians, some researchers have found it useful to define the conflict area within which the associated conflicts occur (4, 7). 1-3

14 One benefit to this approach is that it focuses the analysis of turn-related conflicts and minimizes the potential for non-turn-related conflicts to be included in the conflict count. Figure 1-1 illustrates the location of the pedestrian-vehicle conflict area for left-turning vehicles. It is shown to be as wide as the crosswalk area and as long as the width of the receiving lanes. Opposing lanes 10 ft 10 ft Receiving lanes Pedestrian-vehicle conflict zone Vehicle-vehicle conflict zone Peds - Crosswalk area Subject approach Figure 1-1. Intersection Conflict Areas. Vehicle-Vehicle Conflict Area Left-turning vehicles sometimes conflict with vehicles in the opposing traffic stream. These conflicts can occur when the left-turn vehicle misjudges the adequacy in a gap in the opposing stream. They can also occur when the gap is judged accurately but pedestrian presence in the conflicting crosswalk was not detected until after committing to the left turn. In this situation, the left-turning vehicle stops in the path of the opposing traffic stream, which often causes a conflict with one or more opposing vehicles. These conflicts are relevant to pedestrian safety research because they are indirectly related to pedestrian activity in the crosswalk. Figure 1-1 illustrates the location of the vehicle-vehicle conflict area for left-turning vehicles. It is shown to be as wide as the opposing through traffic lanes and as long as the width of the receiving lanes. Legal Pedestrian A legal pedestrian is considered to be a pedestrian that: (1) enters the crosswalk area during the WALK indication, (2) attempts to enter the crosswalk area during the WALK indication but is prevented from doing so by a turning vehicle, or (3) is present to cross during the WALK indication 1-4

15 but the pedestrian density is so great that the crossing is delayed until the flashing DON T WALK is displayed. REFERENCES 1. Sisiopiku, V., and D. Akin. Pedestrian Behaviors at and Perceptions Towards Various Pedestrian Facilities: An Examination Based on Observation and Survey Data. Transportation Research Part F. Vol. 6. Elsevier Ltd., 2003, p Davis, S., H. Robertson, and L. King. Pedestrian/Vehicle Conflicts: An Accident Prediction Model. Transportation Research Record Transportation Research Board, Washington, D.C., 1989, pp Ragland, D., F. Markowitz, and K. MacLeod. An Intensive Pedestrian Safety Engineering Study Using Computerized Crash Analysis. Institute of Transportation Studies, University of California - Berkeley Traffic Safety Center, Lord, D. Analysis of Pedestrian Conflicts with Left-Turning Traffic. Transportation Research Record Transportation Research Board, Washington, D.C., 1996, pp Carter, D., W. Hunter, C. Zegeer, R. Stewart, and H. Huang. Pedestrian and Bicyclist Intersection Safety Indices: Final Report. Report FHWA-HRT Federal Highway Administration, McLean, Virginia, November Kattan, L., S. Acharjee, and R. Tay. Pedestrian Scramble Operations: Pilot Study in Calgary, Alberta, Canada. Transportation Research Record Transportation Research Board, Washington, D.C., 2009, pp Akin, D., and V. Sisiopiku. Modeling Interactions between Pedestrians and Turning-Vehicles at Signalized Crosswalks Operating Under Combined Pedestrian-Vehicle Interval. Paper No Presented at the 86 th Annual Meeting of the Transportation Research Board, Washington, D.C., January

16

17 CHAPTER 2. LITERATURE REVIEW OVERVIEW For intersections with a permissive or protected-permissive left-turn mode, pedestrians cross during the permissive period. This period occurs during the phase serving opposing through vehicles such that pedestrians cross when the adjacent through vehicles receive a green indication. This operation requires the left-turn driver to yield to both opposing vehicles and pedestrians, prior to accepting a gap and completing the turn. Pedestrian crash risks are increased in these complicated driving conditions because left-turn drivers sometimes fail to yield to the pedestrians. Existing guidelines for the selection of left-turn mode (e.g., permissive, protected, and protected-permissive) focus mainly on vehicular traffic conditions at the intersection. Very few of these guidelines include a sensitivity to pedestrian volume or other pedestrian-safety-related factors. This chapter documents a review of the literature that addresses the topic of pedestrians at signalized intersections. The focus of the review is on research that describes the conflict between pedestrians and left-turning vehicles. There are four parts to this chapter that follow this introductory part. The first part provides some background information on pedestrian safety at intersections in Texas. The second part describes the traffic control elements at signalized intersections that apply to either left-turn movements or pedestrian movements. The third part summarizes research that has quantified pedestrian safety and service quality at intersections. The fourth part summarizes the findings related to traffic engineering treatments that address left-turn-related pedestrian safety at intersections. PEDESTRIAN SAFETY IN TEXAS Data maintained by the National Highway Traffic Safety Administration (NHTSA) indicate that 4654 pedestrian fatalities occurred in the U.S. in 2007, 387 of which occurred in Texas. This number has gradually declined about 1.5 percent each year during the past seven years. NHTSA also estimates that an additional 70,000 pedestrians were injured in 2007, for a total of 74,654 pedestrian involvements. The NHTSA data indicate that 933 pedestrian fatalities occurred at signalized intersections in the U.S. in 2007, 78 of which occurred in Texas. These data indicate that about 20 percent of pedestrian fatalities occur at signalized intersections, which is consistent with the findings reported by Campbell et al. (1) in their analysis of data from the early 1990s. These statistics are summarized in Table 2-1. There are several statistics of note in the data in Table 2-1. The exposure analysis indicates that Texas has a higher rate of pedestrian fatalities and involvements than the U.S. average. A comparison of these rates with those of other states indicates that Texas ranks in the top 16 states. Texas over-representation in these rates extends to signalized intersections. 2-1

18 Table 2-1. Pedestrian Crash Statistics. Location Units Texas U.S. Crash Frequency All Pedestrian fatalities/yr Pedestrian involvements/yr ,654 Vehicle-vehicle crashes/yr 300, Signalized intersections Pedestrian fatalities/yr Exposure Analysis Pedestrian involvements/yr ,509 Exposure Million-vehicle-miles (mvm) All Pedestrian fatalities/mvm Signalized intersections Pedestrian fatalities/mvm Crash Cost Analysis (Willingness-to-Pay Basis) Pedestrian-vehicle crash $200,000/ped. involvement $1.2 billion/yr -- Vehicle-vehicle crash $50,000/veh-veh crash $15 billion/yr -- Note: 1 - Involvements = number of pedestrians that are injured or killed as a result of pedestrian-vehicle crashes. The number of pedestrian involvements each year in Texas (6214) is small compared to the annual number of vehicle-vehicle crashes in Texas (300,000). In fact, pedestrian involvements constitute only about 2 percent of the state s total crashes. However, the severity of a pedestrian crash is much greater than that of a typical vehicle-vehicle crash. A crash cost analysis (based on the willingness-to-pay approach that includes a fatal crash cost estimate of $4 million) indicates that the average cost of a pedestrian involvement is about $200,000, compared to a cost of $50,000 for the average vehicle-vehicle crash. Thus, the estimated cost of pedestrian-related crashes to Texas travelers is $1.2 billion/yr, which is about 8 percent of the total Texas crash cost. INTERSECTION TRAFFIC CONTROL This part of the chapter describes the traffic control elements at signalized intersections that apply to either left-turn movements or pedestrian movements. The first section reviews the various left-turn operational modes and phase sequences used. The second section describes pedestrian signal settings and controller operation to accommodate pedestrian movements. Left-Turn Control Operational Mode The left-turn operational mode is described as protected, protected-permissive, or permissive. A variety of guidelines exist that indicate conditions where the benefits provided by one mode typically outweigh those of other modes. Many of these guidelines indicate that a left-turn phase can be justified based on consideration of several factors that ultimately tie back to the operational or safety benefits derived from the phase. These factors include: 2-2

19 ! Left-turn and opposing through volumes.! Number of opposing through lanes.! Cycle length.! Speed of opposing traffic.! Sight distance.! Crash history. Of note about this list is its lack of a factor that reflects pedestrian presence or flow rate. The flowchart shown in Figure 2-1 presents the guidelines provided in the TxDOT Traffic Signal Operations Handbook (2). They are used to assist in the determination of whether a left-turn phase is needed for a given left-turn movement and whether the operational mode should be permissive, protected, or protected-permissive. These guidelines were derived from a variety of authoritative reference documents (3, 4, 5). The criteria used to determine the operational mode are identified in the various boxes of the flow chart. Guidelines were more recently developed by Yu et al. (6) to address left-turn mode selection. These guidelines include many of the same considerations represented in Figure 2-1. However, there is a notable difference in the thresholds used for the Volume Cross Product check. Phase Sequence Leading, lagging, or split phasing is used when a left-turn phase operates in the protected or protected-permissive mode. The terms leading and lagging indicate the order in which the left-turn phase is presented, relative to the phase serving the conflicting through movement. Leading left-turn phasing has the left-turn phase occurring before the phase serving the conflicting through movement. Lagging left-turn phasing has the left-turn phase occurring last. Split phasing allows all movements on one approach to proceed before those on the opposing approach. For typical intersections, research indicates that lead-lead, lag-lag, and lead-lag phasing provide about the same operational efficiency and safety. Split phasing tends to be less efficient than the other sequences at typical intersections. Thus, the choice between lead-lead, lag-lag, and lead-lag is often based on agency preference or identified benefits to signal coordination. Lag-lag and split phasing are sometimes beneficial at intersections with atypical geometric configurations or volume conditions. 2-3

20 Start Has the critical number of crashes C pt been equalled or exceeded? Yes Protected No Is left-turn driver sight distance to oncoming vehicles less than SD c (equals 5.5 s travel time)? Yes Can sight restriction be removed by offsetting the opposing left-turn lanes? No Protected No Yes How many left-turn lanes are on the subject approach? 2 or more Protected Less than 2 How many through lanes are on the opposing approach? 4 or more Protected Less than 4 Is left-turn volume 2 veh/cycle or less during the peak hour? Yes No Is 85th percentile, or speed limit, of opposing traffic greater than 45 mph? Yes Protected No How many through lanes on the opposing approach? Is V lt x V o > 50,000 during the peak hour? 1 2 or 3 Is V lt x V o > 100,000 during the peak hour? Yes No Yes No Is left-turn delay equal to: a. 2.0 veh-hrs or more, and b. greater than 35 s/veh during the peak hour? Yes No Has the critical number of crashes C p+p been equaled or exceeded? Yes No Permissive Protected + Permissive (desirable) or Protected only Number of Period during Critical Left-Turn-Related Crash Count Left-Turn which Crashes When Considering When Considering Movements on are Considered Protected-only, C pt Prot.+Perm, C p+p Subject Road (years) (crashes/period) (crashes/period) One One One Both Both Both Oncoming Traffic Minimum Sight Distance to Speed Limit (mph) Oncoming Vehicles, SD c (ft) Variables V lt = left-turn volume on the subject approach, veh/h V o = through plus right-turn volume on the approach opposing the subject left-turn movement, veh/h Figure 2-1. Guidelines for Determining Left-Turn Mode. (2) 2-4

21 Lead-Lead Left-Turn Phasing. The most commonly used left-turn phase sequence is the lead-lead sequence, which has both opposing left-turn phases starting at the same time. The advantages of this phasing option are:! It is consistent with driver expectation such that drivers react quickly to the leading green arrow indication.! It minimizes conflicts between left-turn and through vehicles on opposing approaches by clearing left-turn vehicles first and, thereby, reducing the number of left-turn drivers that must find safe gaps.! It minimizes conflicts between left-turn and through movements on the same approach when the left-turn volume exceeds its available storage length. Lag-Lag Left-Turn Phasing. This left-turn phase sequence has both opposing left-turn phases ending at the same time. The advantages of the lag-lag phasing option are:! It ensures that both adjacent through phases start at the same time a characteristic that is particularly amenable to efficient signal coordination with pretimed control.! If used with the protected-permissive mode, it minimizes presentation of the left-turn phase during low-volume conditions by clearing left-turn vehicles during the initial through phase.! If used with the protected-permissive mode for the major road as part of a coordinated signal system, then it reduces delay to major-road left-turn movements by serving them soon after arrival. The yellow trap problem is created when lag-lag phasing is used for opposing left-turn movements, and the left-turn phases operate in the protected-permissive mode. Flashing-yellowarrow operation is one technique for eliminating this problem. It retains a permissive indication for left-turn drivers during the change period for the adjacent through movement phase (7). Lead-Lag Left-Turn Phasing. The lead-lag left-turn phase sequence is sometimes used to accommodate through movement progression in a coordinated signal system. The aforementioned yellow trap may occur if the leading left-turn movement operates in the protected-permissive mode and the two through movement phases time concurrently during a portion of the cycle. This sequence may also be used at intersections where the leading left-turn movement is not provided an exclusive storage bay (or a bay is provided but it does not have adequate storage). Split Phasing. Split phasing refers to the sequential service of opposing intersection approaches in two phases. This phasing is typically less efficient than lead-lead, lead-lag, or lag-lag left-turn phasing. It often increases the cycle length, or if the cycle length is fixed, it reduces the time available to the intersecting road. Split phasing may be helpful if there is a need to serve left turns from the opposing approaches, but sufficient width is not available to ensure their adequate separation in the middle of the intersection if served concurrently. 2-5

22 Pedestrian Control This section describes pedestrian signal settings and controller operation to accommodate pedestrian movements. Specific topics of discussion include pedestrian signal heads, pedestrian intervals, leading pedestrian interval, and exclusive pedestrian phase. Pedestrian Signal Heads The Texas Manual on Uniform Traffic Control Devices (TMUTCD) indicates that pedestrian signal heads should be used under any of the following conditions:! If it is necessary to assist pedestrians in making a reasonably safe crossing or if engineering judgment determines that pedestrian signal heads are justified to minimize vehicle-pedestrian conflicts;! If pedestrians are permitted to cross a portion of a street, such as to or from a median of sufficient width for pedestrians to wait, during a particular interval but are not permitted to cross the remainder of the street during any part of the same interval; and/or! If no vehicular signal indications are visible to pedestrians, or if the vehicular signal indications that are visible to pedestrians starting or continuing a crossing provide insufficient guidance for them to decide when it is reasonably safe to cross, such as on one-way streets, at T-intersections, or at multiphase signal operations (8). The Manual on Uniform Traffic Control Devices (MUTCD) includes additional guidance on the use of countdown indications in conjunction with the pedestrian signal. Specifically, it states, All pedestrian signal heads used at crosswalks where the pedestrian change interval is more than 7 seconds shall include a pedestrian change interval countdown display... (9). Pedestrian Intervals This section provides guidelines for determining the duration of the walk interval and the pedestrian change (i.e., pedestrian clear) interval. Walk Interval. The walk interval gives pedestrians adequate time to perceive the WALK indication and depart the curb before the pedestrian change interval begins. The TMUTCD indicates that the minimum walk duration should be at least 7 s, but indicates that a duration as low as 4 s may be used if pedestrian volume is low or pedestrian behavior does not justify the need for 7 s (8). Consideration should be given to longer walk duration in school zones and areas with large numbers of older pedestrians. Pedestrian Change Interval. Pedestrian clearance time must follow the walk interval. It should allow a pedestrian crossing in the crosswalk to leave the curb (or shoulder) and walk at a normal rate to at least the far side of the traveled way, or to a median of sufficient width for pedestrians to wait ( ). The TMUTCD ( ) recommends a walking speed value of 4.0 ft/s. However, Fitzpatrick et al. (10) recommend a maximum walking speed of 3.5 ft/s for general pedestrian populations and 2-6

23 3.0 ft/s if older pedestrians are a concern. The MUTCD indicates that a walking speed of 3.5 ft/s should be used (9). However, it also indicates that a walking speed of up to 4.0 ft/s may be used to evaluate the sufficiency of the pedestrian clearance time at locations where an extended push button press function has been installed to provide slower pedestrians an opportunity to request a longer pedestrian clearance time (9). The pedestrian clearance time is computed as the crossing distance divided by the walking speed. Crossing distance is typically measured from curb to curb along the crosswalk. Clearance time can be obtained from Table 2-2 for typical pedestrian crossing distances and walking speeds. Table 2-2. Pedestrian Clearance Time. Pedestrian Crossing Walking Speed, ft/s Distance, ft Pedestrian Clearance Time (PCT), s Note: 1 - Clearance times computed as PCT = D c / v p, where D c = pedestrian crossing distance (in feet) and v p = pedestrian walking speed (in feet per second). The TMUTCD indicates that the pedestrian clearance time can be provided during: (1) the pedestrian change interval during which a flashing DON T WALK indication is displayed, and (2) a second interval that times concurrent with the vehicular yellow change and red clearance intervals and displays either a flashing or steady DON T WALK indication (8). This practice minimizes the impact of pedestrian service on phase duration and allows the phase to be responsive to vehicular demand. Following this guidance, the pedestrian change interval (also called the pedestrian clear interval) is computed using Equation 1. where, PCI = pedestrian change interval duration, s. PCT = pedestrian clearance time, s. Y = yellow change interval, s. R c = red clearance interval, s. (1) 2-7

24 Leading Pedestrian Interval The leading pedestrian interval is a feature available in most modern traffic controllers. It displays the WALK indication a few seconds before the green ball indication. In this manner the pedestrians can establish a presence in the crosswalk before turning vehicles receive the permissive green indication. A lagging pedestrian interval is also available in most controllers, but it gives preference to vehicular traffic by allowing them to start before the pedestrians. Exclusive Pedestrian Phase An exclusive pedestrian phase serves all pedestrian crossing movements simultaneously, while holding all vehicular movements with a red signal indication. The minimum duration of the phase is sufficient to allow the pedestrian to react to the WALK indication and cross one intersection leg, perpendicular to the direction of traffic flow. This type of phasing is occasionally used in the central business district of large cities. A variation of the exclusive pedestrian phase is the pedestrian scramble. It permits pedestrians to cross on a diagonal path through the intersection conflict area. It also allows the more traditional, perpendicular crossing movements across an intersection leg. The timing of this phase is sufficient to allow the pedestrian to react and cross the longest diagonal path. LEFT-TURN-RELATED SAFETY ISSUES This part of the chapter summarizes the research that has quantified intersection pedestrian service quality and safety, as influenced by left-turning vehicles. Initially, a model is described for estimating the pedestrian s perception of service quality provided at an intersection. Then, a series of safety prediction models are examined. These models predict the frequency of pedestrian crashes at intersections. Next, the effect of left-turn control on pedestrian safety is examined. Finally, alternative pedestrian controls are examined in the context of their reported ability to improve pedestrian safety. Pedestrian Service at Signalized Intersections Dowling et al. (11) developed procedures for quantifying the level of service provided to pedestrians at signalized intersections. The procedure consists of a series of calculations that focus on quantifying the service provided by one crosswalk at the intersection; it is repeated as needed for each crosswalk of interest. The procedure is sensitive to the volume of right-turning and left-turning vehicles crossing the crosswalk. It also considers the delay incurred by pedestrians and the speed of the through traffic stream. Output from the procedure is a numeric score ranging from 1 to 6. Lower values indicate a very good level of service. In fact, values less than or equal to 2.0 represent level-of-service A, values between 2.0 and 2.75 represent level-of-service B, and so on. The procedure was used to evaluate the sensitivity of pedestrian level of service to traffic volume and turn percentage. The intersection used for the evaluation had four legs and two lanes on each approach. Two volume scenarios were evaluated. In one scenario, the turn movements were each 10 percent of the approach volume. In the other scenario, turn movements were each 20 percent 2-8

25 of the approach volume. The major-street volume equaled the minor-street volume for both scenarios. Each phase duration was computed to be in proportion to its flow ratio. The results of the evaluation are shown in Figure 2-2. Pedestrian Level of Service Minor-street volume = major-street volume E 20% turns D C Left and right turns = 10% of approach volume B A Major-Street Approach Volume, veh/h Figure 2-2. Pedestrian Level of Service Based on Traffic Volume. The trends in Figure 2-2 reflect the analysis of one crosswalk at a signalized intersection. The numeric scores indicating level of service are reflected on the y-axis on the left side of the figure. The corresponding level-of-service letter is indicated on the right side of the figure. The two trend lines indicate that pedestrian level of service degrades as approach volume increases. The increase in turn percentage is also shown to have a negative effect on service quality. It reflects the increased conflict between turning vehicles and pedestrians when using the crosswalk. Pedestrian Safety at Signalized Intersections Lyon and Persaud (12) examined pedestrian crash frequency for signalized intersections in Toronto, Canada. They gathered pedestrian volume and vehicle volume data for 684 four-leg intersections and 263 three-leg intersections. Eleven years of pedestrian crash data were obtained for each intersection. Lyon and Persaud used the data to calibrate one crash prediction model for four-leg intersections and a second model for three-leg intersections. Both models predicted the expected annual pedestrian crash frequency at an intersection. They included a sensitivity to the total daily entering volume, count of pedestrians using each sidewalk during an eight-hour period (totaled for all crosswalks), and total daily left-turning volume entering the intersection. It is noted that the models predict the annual number of reported crashes for an intersection even though the input pedestrian volume represents only eight hours of the day. 2-9

26 The models calibrated by Lyon and Persaud were used to evaluate the sensitivity of pedestrian crash frequency to daily vehicular volume and left-turn percentage. For four-leg intersections, the major-street volume equaled the minor-street volume. For three-leg intersections, the minor-street volume equaled one-half of the major-street volume. The eight-hour pedestrian volume was based on an assumed 50 p/h in each crosswalk during each of the eight hours. The results of the evaluation are shown in Figure 2-3. Pedestrian Crash Frequency, cr/yr/intersection legs: minor-street volume = 0.5 major-street volume 4-legs: minor-street volume = major-street volume 50 p/h crossing each leg for 8 hours 4-legs, 20% left turns 4-legs, 10% 3-legs, 20% left turns 3-legs, 10% Major-Street Daily Volume, veh/d Figure 2-3. Crash Frequency Based on Traffic Volume. The trend lines in Figure 2-3 indicate that pedestrian crash frequency increases with an increase in vehicular volume. A four-leg intersection with an average daily volume of 20,000 veh/d on each street, 20 percent left turns, and 20 percent right turns is likely to have one reported pedestrian crash each year. The four-leg intersections have more than twice as many crashes as the three-leg intersections for a given volume and turn percentage. A 15 percent increase in daily volume corresponds to about an 8 percent increase in crashes. In contrast, a 15 percent increase in left-turn volume corresponds to a 4 percent increase in crashes. These trends suggest that left-turn volume and through volume do not increase crash risk (i.e., that intersections with higher volume have a lower crash rate). It is possible that, once the first left-turn vehicle initiates its maneuver safely, the remaining left-turn vehicles also proceed safely regardless of their number. It follows then, that crash risk may increase more notably with an increase in the number of signal cycles per hour. 2-10

27 Effect of Left-Turn Control on Pedestrian Safety Lord (13) reviewed several research reports on the topic of left-turn-related pedestrianvehicle crashes. He found that left-turn maneuvers account for 20 to 30 percent of all pedestrian crashes at intersections. Left-turn-related pedestrian crashes are exceeded in number only by collisions between pedestrians and through vehicles (51 percent). Research indicates that left-turn mode (i.e., permissive, protected-permissive, or protected) and phase sequence have some influence on the safety of the pedestrian crossing maneuver. Each of these influences is discussed in the following subsections. Permissive Left-Turn Mode A driver turning left at a signalized intersection during a permissive period has to monitor multiple information sources during the maneuver. These sources include the signal indication, opposing vehicle stream, and pedestrian activity in the crosswalk that is adjacent to the opposing through traffic lanes. These multiple sources make the left-turn maneuver relatively complex and place a significant demand on the driver workload. Sometimes, when this workload exceeds the driver s processing capability, an incorrect decision is made and a left-turn-related crash occurs. Quaye et al. (14) examined left-turn-related pedestrian crash frequency for individual crosswalks as a function of the pedestrian volume and left-turn volume in the crosswalk. They collected pedestrian and vehicular volume data for 547 crosswalks at 200 intersections in Canada. They combined this data with left-turn-related crash data for a four-year period. The resulting database was used to develop a model for predicting the expected annual number of left-turn-related pedestrian crashes during a specified hour of the day. Thus, the model is applied to each of the 24 hours during a representative day, and the 24 estimates are then added to obtain an estimate of the annual crash frequency for the subject crosswalk. One of the models calibrated by Quaye et al. applies to crosswalks where the conflicting leftturn movement operated in the permissive mode and was opposed by a through vehicle traffic movement. The predicted relationship between left-turn-related pedestrian crash frequency, left-turn volume, and pedestrian volume is shown in Figure 2-4a. The trends in Figure 2-4a are consistent with those in Figure 2-3 and indicate that crash frequency increases with increasing left-turn volume. A 15 percent increase in left-turn volume corresponds to a 5 percent increase in crash frequency. This trend suggests that left-turn volume does not increase crash risk a trend that is consistent with that observed in Figure 2-3. A 15 percent increase in pedestrian volume corresponds to a 12 percent increase in crash frequency. The nearly one-to-one ratio of these percentages suggests that pedestrian volume has negligible effect on risk (i.e., crash rate is uninfluenced by pedestrian volume). The trends in Figure 2-4a also suggest that there is one pedestrian crash every 6 to 12 years for a given crosswalk, when the conflicting left-turn volume ranges from 100 to 150 veh/h. 2-11

28 Pedestrian Crash Frequency, cr/yr/crosswalk p/h crossing each leg for 8 hours; 30 p/h thereafter 50 p/h crossing each leg for 8 hours; 15 p/h thereafter Opposing vehicular volume, permissive-only left-turn operation Average Left-Turn Volume, veh/h Pedestrian-Vehicle Conflicts, conflicts/day/crosswalk Opposing vehicular volume, permissive-only left-turn operation 100 p/h crossing each leg for 8 hours; 30 p/h thereafter 50 p/h crossing each leg for 8 hours; 15 p/h thereafter Average Left-Turn Volume, veh/h a. Crash Frequency. b. Conflict Frequency. Figure 2-4. Crash and Conflict Frequency for Permissive Operation. Akin and Sisiopiku (15) collected pedestrian-vehicle conflict data at three signalized intersections in Michigan. One crosswalk at each intersection was videotaped for a period of three to four hours during one day. Pedestrian volume, left-turn volume, and pedestrian-vehicle conflicts were extracted from the videotape for each 30-minute time period. Two of the intersections had permitted left-turn operation with an opposing through vehicle movement. Their examination of the data indicated that hourly conflict frequency was linearly related to the product of the hourly leftturn volume and the hourly pedestrian volume. The linear relationship shown in Figure 2-4b is derived by the authors of this report using data reported by Akin and Sisiopiku. The trends in Figure 2-4b indicate that the number of conflicts in a 24-hour period is roughly equal to the left-turn volume. More specifically, when the left-turn volume averages 100 veh/h, the pedestrian volume averages 50 p/h during eight hours, and 15 p/h during the remaining hours, then about 60 conflicts occur each day. If the pedestrian volume doubles, then about 120 conflicts occur each day. A relationship between conflict and crash frequency can be derived by combining the functions represented in Figure 2-4. This relationship is shown in Figure 2-5. The trends in this figure give an indication of the large number of conflicts that occur relative to the number of crashes. Roughly speaking, if a crosswalk is found to experience conflicts at a rate of 100 per day, then it will experience one crash every 7 to 10 years. Unopposed Left-Turn Operation In addition to examining left-turn-related pedestrian crash frequency at intersections with permitted left-turn operation, Quaye et al. (14) also examined crash frequency at intersections where there was no opposing through vehicle movement. This condition is found at: (1) intersections where one or both intersecting streets serve only one travel direction and (2) three-leg intersections. The left turns from a one-way leg or from the terminating leg of a three-leg intersection do not have an opposing through vehicle movement. 2-12

29 Pedestrian Crash Frequency, cr/yr/crosswalk p/h crossing each leg for 8 hours; 30 p/h thereafter 50 p/h crossing each leg for 8 hours; 15 p/h thereafter Opposing vehicular volume, permissive-only left-turn operation Pedestrian-Vehicle Conflicts, conflicts/day/crosswalk Figure 2-5. Crash and Conflict Relationship for Permissive Operation. Quaye et al. developed a pedestrian-vehicle crash prediction model for crosswalks with unopposed left-turn operation. The relationship between left-turn-related pedestrian crash frequency, left-turn volume, and pedestrian volume is shown in Figure 2-6a. Pedestrian Crash Frequency, cr/yr/crosswalk p/h crossing each leg for 8 hours; 30 p/h thereafter Average Left-Turn Volume, veh/h Pedestrian-Vehicle Conflicts, conflicts/day/crosswalk 50 p/h crossing each leg for 8 hours; 50 p/h crossing each leg for 8 hours; 15 p/h thereafter 15 p/h thereafter No opposing vehicular volume (e.g., 3-leg intersection) No opposing vehicular volume (e.g., 3-leg int.) 100 p/h crossing each leg for 8 hours; 30 p/h thereafter Average Left-Turn Volume, veh/h a. Crash Frequency. b. Conflict Frequency. Figure 2-6. Crash and Conflict Frequency for Unopposed Left-Turn Operation. The trends in Figure 2-6a are not fully consistent with those in Figures 2-3 or 2-4a. They indicate that crash frequency increases rapidly with increasing left-turn volume. This finding suggests that pedestrians are less safe in crosswalks where the conflicting left-turn volume is high and the left-turn movement does not have an opposing through vehicle movement to partially shield pedestrians at the start of the phase. 2-13

30 More specifically, the trend lines in Figure 2-6a indicate that a 15 percent increase in leftturn volume corresponds to a 20 percent increase in crash frequency. This trend suggests that increasing left-turn volume increases crash risk. On the other hand, a 15 percent increase in pedestrian volume corresponds to a 4 percent increase in crash frequency. This trend suggests that risk decreases with increasing pedestrian volume a trend that was also noted by Leden (16) in a subsequent re-examination of the Quaye et al. data. The trends in Figure 2-6a also suggest that there is one pedestrian crash every 3 to 8 years for a given crosswalk, when the conflicting left-turn volume ranges from 100 to 150 veh/h. One of the intersections studied by Akin and Sisiopiku (15) was a three-leg intersection. They collected conflict data on the left turn from the terminating leg. The linear relationship is derived by the authors of this report using the data reported by Akin and Sisiopiku. It is shown in Figure 2-6b. The trends in Figure 2-6b indicate that the number of conflicts in a 24-hour period is roughly equal to twice the left-turn volume. More specifically, when the left-turn volume averages 100 veh/h, the pedestrian volume averages 50 p/h during eight hours, and 15 p/h during the remaining hours, then about 155 conflicts occur each day. If the pedestrian volume doubles, then about 310 conflicts occur each day. A relationship between conflict and crash frequency was derived by combining the functions represented in Figure 2-6. This relationship is shown in Figure 2-7. The trend lines in this figure illustrate the large number of conflicts that occur relative to the number of crashes. A comparison between Figures 2-5 and 2-7 indicates that, while conflicts are more frequent at left-turn locations with unopposed left-turn operation, they tend to result in fewer crashes per year for typical pedestrian and vehicle volume combinations. Roughly speaking, if a crosswalk is found to experience conflicts at a rate of 100 per day, then it will experience only one crash every 13 to 29 years. Protected-Permissive Left-Turn Mode The relationship between pedestrian safety and the protected-permissive mode has not been established through research. It could be rationalized that protected-permissive operation is less safe than a protected left-turn operation and more safe than permissive operation, given that it combines the operation of both modes. A flashing yellow arrow implementation of protected-permissive operation was researched by Brehmer et al. (7) in terms of its safety and operational benefits. They found that the flashing yellow arrow indication has a lower fail-critical rate as compared to the circular green indication when used to indicate the permissive period. A fail-critical response occurs when the left-turning driver incorrectly interprets the permissive indication as a protected turn indication, thus creating the potential for a crash with opposing vehicles or pedestrians. By inference, the flashing-yellowarrow display would offer some benefit to pedestrian safety through better driver comprehension of the signal display; however, there is no research to confirm this inference. 2-14

31 Pedestrian Crash Frequency, cr/yr/crosswalk p/h crossing each leg for 8 hours;15 p/h thereafter 100 p/h crossing each leg for 8 hours; 30 p/h thereafter No opposing vehicular volume (e.g., 3-leg intersection) Pedestrian-Vehicle Conflicts, conflicts/day/crosswalk Figure 2-7. Crash and Conflict Relationship for Unopposed Left-Turn Operation. Koonce et al. (17) point out the use of the permissive period omit as a technique to improve pedestrian safety. With this technique, the controller is set up to omit the permissive period during any cycle in which a detection is received for a conflicting pedestrian movement. However, this technique has the possible disadvantage of violating driver expectancy. One issue that is related to the use of the protected-permissive mode is its possible negative impact on pedestrian compliance with the pedestrian signal. One factor that affects pedestrian compliance is waiting time, which is typically increased when permissive operation is replaced by protected-permissive operation. Studies indicate that pedestrian compliance degrades with waiting time. Pedestrians are reluctant to wait more than 30 s, and compliance is notably poor if the wait is 60 s or more (18). Protected Left-Turn Mode Guidance on p. 457 of the Traffic Control Devices Handbook (19) indicates that the choice of left-turn mode should be based on consideration of the overall safety and efficiency of the intersection. They caution that protected-only left-turn phases may improve the safety of leftturning vehicles, but this will be accomplished at the expense of other movements and pedestrians. The implied expense is an increase in delay to vehicles and pedestrians. One additional concern is whether the increase in pedestrian delay will lead to a decrease in pedestrian compliance. Leading vs. Lagging Phase Sequence Hummer et al. (20) measured conflicts between pedestrians and left-turn vehicles at signalized intersections. Some of the intersections they evaluated had a leading left-turn phase. Other intersections had a lagging left-turn phase. They found that the leading left-turn phase was associated with six times as many conflicts as the lagging left-turn phase. In most instances, the 2-15

32 leading-left-turn-related conflicts were a result of the pedestrian mistaking the end of the cross street through phase for the start of the through phase in the direction they were traveling (when, in fact, it was the start of the left-turn phase). The pedestrians would not see, or disregard, the DON T WALK indication and step into the crosswalk and the path of the left-turn vehicle. In contrast, the lagging sequence not only meets pedestrian expectations, but its through-vehicle queue provides a natural left-turn vehicle shield for pedestrians at the start of the permissive period. Effect of Pedestrian Control on Pedestrian Safety Research indicates that there are several traffic control techniques that can be used to address left-turn-related pedestrian-vehicle conflicts and crashes. These techniques include: leading pedestrian interval, exclusive pedestrian phase, pedestrian clear interval, and selected warning signs or markings. Each of these techniques is described in the following subsections. Leading Pedestrian Interval Lalani (21) indicates that agencies have typically implemented a leading pedestrian interval at locations where 3 to 20 percent of turning vehicles typically violate the pedestrian right-of-way. The leading interval always increases vehicle delay (22) so its safety benefit must outweigh its adverse impact on vehicle operation. A lagging pedestrian interval is available in most controllers but it does not reduce left-turn-related pedestrian conflicts, and may even increase them. Fayish and Gross (23) examined pedestrian crash data at 10 signalized intersections in State College, Pennsylvania, at which a leading pedestrian interval was installed. For each intersection, they gathered crash data for four years before the leading pedestrian interval was installed and for three years after installation. The leading pedestrian interval was 3.0 s in duration at each intersection. Their analysis indicated that pedestrian-vehicle crash frequency at intersections decreased by 58.7 percent after the leading pedestrian intervals were implemented. This reduction includes pedestrian crashes associated with both left-turning and right-turning vehicles. Exclusive Pedestrian Phase An exclusive pedestrian phase provides temporal separation between vehicles and pedestrians by providing each travel mode exclusive use of the intersection during a signal phase. Research by Zegeer et al. (24) found that this phasing arrangement is associated with a 50 percent reduction in pedestrian crashes, relative to signalized intersections with concurrent service of pedestrians during the through vehicle phases and intersections with no pedestrian signals. A scramble phase is a special type of exclusive pedestrian phase. It was also found to provide a similar safety benefit as the exclusive pedestrian phase; however, it is reported to work best when pedestrian volume exceeds 1200 pedestrians per day, street widths are narrow (e.g., less than 60 ft), and through movement volume is low (22, 24). 2-16

33 Provide Pedestrian Clear Entirely during Green Some agencies prefer to minimize the duration of the pedestrian change interval by subtracting the yellow change interval and red clearance interval (see Equation 1). However, this practice may cause some conflict between pedestrians and left-turning vehicles that are clearing the intersection following the permissive portion of the phase. A similar conflict can occur if permissive or protected-permissive left-turn operation is used with the rest-in-walk mode. If permissive or protected-permissive left-turn operation is used and vehicular volume is low enough that the phase ends after timing the pedestrian walk and change intervals, then the pedestrian change interval (also known as pedestrian clear interval) can be set to equal the pedestrian clearance time, as defined in the text associated with Table 2-2. Under these conditions, the pedestrian clear times entirely during the green interval, and some vehicle-pedestrian conflicts that occur at the end of the phase may be alleviated. Turn Vehicle Warning Signs and Markings Signs and/or markings with a message such as, Pedestrians Watch for Turning Vehicles have been used to increase the awareness of pedestrians at intersections. The marking is placed in the crosswalk near the curb. The sign is placed on the far-side pole facing pedestrians in the crosswalk. Research indicates that these treatments reduce left-turn-related pedestrian-vehicle conflicts by 20 to 60 percent (21). A sign with the message, Yield to Pedestrians When Turning was evaluated by Zegeer et al. (24) using a before-after study. They installed the sign at four intersections and used it to inform turning drivers of pedestrian presence. They used vehicle-pedestrian conflicts as the measure of effectiveness. They found that the sign had no significant effect on left-turn-related conflict frequency, but total pedestrian-vehicle conflicts were reduced 25 to 37 percent. Effect of Pedestrian Flow on Vehicle Operation Chapter 16 of the 2000 Highway Capacity Manual (25) describes a methodology for evaluating vehicle operation at a signalized intersection. The methodology is sensitive to pedestrian volume. It models the effect of pedestrian presence on discharging left- and right-turn vehicles by decreasing the vehicle s saturation flow rate. Higher pedestrian volume in a crosswalk corresponds to lower saturation flow rate and a higher delay. The Chapter 16 methodology was used to examine the sensitivity of delay to vehicle and pedestrian volume. A four-leg intersection was devised for this examination. Each leg served twoway traffic and provided two lanes in each travel direction. No turn bays or phases were provided. Vehicular demand on each street was the same, 20 percent of the approach traffic turned left, and another 20 percent turned right. A pedestrian volume of 400 p/h was used, which is typical of an intersection in a central business district (25). The results are shown in Figure

34 Major-Street Control Delay, s/veh Minor-street volume = major-street volume Major-street left- and right-rurns: 20% Minor-street left- and right-turns: 20% 2 lanes on each approach, no turn bays 2 phases 400 p/h Major-Street Approach Volume, veh/h 0 p/h Figure 2-8. Relationship between Pedestrian Volume and Vehicle Delay. The trend lines in Figure 2-8 indicate that pedestrian volume has a minimal effect on vehicle delay when approach volumes are less than 600 veh/h (i.e., 300 veh/h/ln). As the volume increases above 600 veh/h, a pedestrian volume of 400 p/h is shown to increase delay by several seconds per vehicle. In fact, at a vehicular volume of 800 veh/h, the increase in delay due to pedestrian presence is about 10 s/veh. CANDIDATE TRAFFIC ENGINEERING TREATMENTS A review of the literature indicates a range of traffic engineering treatments are available to address left-turn-related pedestrian-vehicle conflicts. The use of the protected or protectedpermissive mode is a commonly cited treatment. However, there were instances where other treatments were found to provide some safety benefit without increasing vehicle delay or degrading progression quality as much as the addition of a left-turn phase. These treatments are summarized in Table 2-3. The effectiveness of each treatment may vary, depending on whether the subject leftturn movement is opposed by a through vehicular movement. 2-18

35 Table 2-3. Traffic Engineering Treatments for Improving Pedestrian Safety at Intersections. Treatment Description Issues Treatments Based on Conversion from Permissive Mode Provide protected-permissive mode Provide protected-permissive mode using flashing yellow signal display Reduce the number of left-turn vehicles that turn during the permissive period by providing a protected arrow indication. Reduce the number of left-turn vehicles that turn during the permissive period by providing a protected arrow indication. Permissive period presents opportunity for some pedestrian-vehicle conflicts. If used with lagging left-turn phase sequence, safety problems associated with the yellow trap may occur. Safety benefits of protected-permissive phasing with flashing yellow display not quantified through research. Provide protected left-turn mode (19, 22) Provide lagging phase sequence (20) Separate vehicles and pedestrians on problem approach by providing each a separate time in cycle to be served. Through phase (and pedestrian service) occur after cross street phase ends and is consistent with pedestrian expectation. Treatments Used in Conjunction with Permissive Mode Provide leading pedestrian interval (19, 21, 22) Provide exclusive pedestrian phase (19,22) Provide pedestrian clear interval entirely during green (2) Add turning vehicle warning signs and markings (19, 21) Prohibit pedestrian crossing Provide a small amount of time (say, 3 s) to allow pedestrians to start crossing before displaying green ball. Separate vehicles and pedestrians at intersection by providing each a separate time in cycle to be served. Some agencies time a portion of the pedestrian clearance time during the yellow change and red clearance intervals to reduce delay. Add Pedestrians Watch for Turning Vehicles signs and/or markings to remind pedestrians to look for turning vehicles. Redirect pedestrians to alternative crosswalks or crossing locations. Left-turn phase may increase pedestrian waiting time and decrease their compliance with the pedestrian signal. Treatment is viable with protected mode, and with protected-permissive mode provided that yellow trap is eliminated. Increases delay to vehicles. May require the use of accessible pedestrian signals to provide crossing cues to visually impaired pedestrians. Significant delay and waiting time may result if used at large intersections. May require the use of accessible pedestrian signals to provide crossing cues to visually impaired pedestrians. Increases delay to vehicles. May only provide pedestrian safety benefit if used with protected left-turn mode or with permissive omit when ped. is detected. Markings tend to wear away when used in typical crosswalk locations. May shift safety problem to another intersection. May cause negative public reaction unless it is part of a larger traffic management strategy. REFERENCES 1. Campbell, B., C. Zegeer, H. Huang, and M. Cynecki. A Review of Pedestrian Safety Research in the United States and Abroad. Report No. FHWA-RD Federal Highway Administration, U.S. Department of Transportation, Washington, D.C., January

36 2. Bonneson, J., S. Sunkari, and M. Pratt. Traffic Signal Operations Handbook. Report No. FHWA/TX-09/ P1. Texas Transportation Institute, College Station, Texas, March Kell, J.H., and I.J. Fullerton. Manual of Traffic Signal Design. 2nd Edition. Institute of Transportation Engineers, Washington, D.C., Orcutt, F. The Traffic Signal Book. Prentice-Hall, Inc., Englewood Cliffs, New Jersey, Traffic Engineering Handbook. 5th Edition. J.L. Pline editor. Institute of Transportation Engineers, Washington, D.C., Yu, L., Y. Qi, H. Yu, L. Guo, and X. Chen. Development of Left-Turn Operations Guidelines at Signalized Intersections. Report No. FHWA/TX-09/ Texas Southern University, Houston, Texas, March Brehmer, C., K. Kacir, D. Noyce, and M. Manser. NCHRP Report 493: Evaluation of Traffic Signal Displays for Protected/Permissive Left-Turn Control. TRB, National Research Council, Washington, D.C., Texas MUTCD: Manual on Uniform Traffic Control Devices. Texas Department of Transportation, Austin, Texas, Manual on Uniform Traffic Control Devices. Federal Highway Administration, U.S. Department of Transportation, Washington, D.C., Fitzpatrick, K., S. Turner, M. Brewer, P. Carlson, B. Ullman, N. Trout, E. Park, and J. Whitacre. NCHRP Report 562: Improving Pedestrian Safety at Unsignalized Crossings. TRB, National Research Council, Washington, D.C., Dowling, R., D. Reinke, A. Flannery, P. Ryus, M. Vandehey, T. Petritsch, B. Landis, N. Rouphail, and J. Bonneson. NCHRP Report 616: Multimodal Level of Service Analysis for Urban Streets. NCHRP, National Research Council, Washington, D.C., Lyon, C., and B. Persaud. Pedestrian Collision Prediction Models for Urban Intersections. Transportation Research Record Transportation Research Board, Washington, D.C., 2002, pp Lord, D. Analysis of Pedestrian Conflicts with Left-Turning Traffic. Transportation Research Record Transportation Research Board, Washington, D.C., 1996, pp Quaye, K., L. Leden, and E. Hauer. Pedestrian Accidents and Left-Turning Traffic at Signalized Intersections. AAA Foundation for Traffic Safety, Washington, D.C., April

37 15. Akin, D., and V. Sisiopiku. Modeling Interactions between Pedestrians and Turning-Vehicles at Signalized Crosswalks Operating Under Combined Pedestrian-Vehicle Interval. Paper No Presented at the 86 th Annual Meeting of the Transportation Research Board, Washington, D.C., January Leden, L. Pedestrian Risk Decrease with Pedestrian Flow. A Case Study Based on Data from Signalized Intersections in Hamilton, Ontario. Accident Analysis & Prevention. No. 34, pp Koonce, P., L. Rodegerdts, K. Lee, S. Quayle, S. Beaird, C. Braud, J. Bonneson, P. Tarnoff, and T. Urbanik. Traffic Signal Timing Manual. Report No. FHWA-HOP Federal Highway Administration, U.S. Department of Transportation, Washington, D.C., June Neudorff, L. Candidate Signal Warrants from Gap Data - Technical Report. Report No. FHWA-RD-82/152. Federal Highway Administration, U.S. Department of Transportation, Traffic Control Devices Handbook. Chapter 13 - Pedestrians. Institute of Transportation Engineers, Washington, D.C., Hummer, J., R. Montgomery, and K. Sinha. Guidelines for Use of Leading and Lagging Left- Turn Signal Phasing. Transportation Research Record Transportation Research Board, National Research Council, Washington, D.C., 1991, pp Lalani, N. Alternative Treatments for At-Grade Pedestrian Crossings. Institute of Transportation Engineers, Washington, D.C., Traffic Engineering Technical Committee TENC-5A-5. Design and Safety of Pedestrian Facilities. (C. Zegeer, Chair). Institute of Transportation Engineers, Washington, D.C., March Fayish, A., and F. Gross. Safety Effectiveness of Leading Pedestrian Intervals Evaluated by a Before-After Study with Comparison Groups. Transportation Research Record Transportation Research Board, Washington, D.C., 2010, pp Zegeer, C., K. Opiela, and M. Cynecki. Pedestrian Signalization Alternatives. Report No. FHWA/RD-83/102. Federal Highway Administration, U.S. Department of Transportation, Highway Capacity Manual. Transportation Research Board, National Research Council, Washington, D.C.,

38

39 CHAPTER 3. STATE OF THE PRACTICE REVIEW OVERVIEW This chapter documents the findings from a survey of practitioners regarding the state-of-thepractice in addressing pedestrian safety at signalized intersections. It consists of two parts. The first part documents the information obtained during interviews with practitioners in Texas and several other states. The second part documents the findings from an electronic discussion (e-discussion) that was conducted with an expert panel comprised of 12 state and city traffic engineers. PRACTITIONER INTERVIEW This part of the chapter documents the information obtained during interviews with traffic engineers responsible for signalized intersection operations in various cities and states. The interview addressed pedestrian safety at signalized intersections in urban areas, with a focus on conflicts between left-turn vehicles and pedestrians. The interview objectives were as follows:! Assess the state of the practice regarding control for left turns and pedestrians.! Determine how pedestrian needs are assessed and accommodated.! Identify treatments used to address left-turn-related pedestrian safety concerns. In addition to providing the aforementioned information, the interviews were also used to identify people who could serve on an expert panel. The expert panel was used to conduct an indepth exploration of pedestrian safety problems and potential treatments. The findings from this activity are documented in the next part of this chapter. Background The researchers conducted interviews with practitioners in several TxDOT districts, Texas cities, and public agencies in other states. The interviewees were asked questions about their left-turn signalization practices, methods for accommodating pedestrian needs, and use of treatments to address pedestrian safety. The Texas interviewees were also asked to assist in identifying candidate field data collection sites. The specific questions asked during the interviews are listed in Figures 3-1 and 3-2. The responses to the interview questions are documented in the following subsections. 3-1

40 INTERVIEW QUESTIONS Inventory and Practice 1. How many urban signalized intersections do you operate in your jurisdiction? 2. What percentage of these intersections have crosswalks... a. On all approach legs? b. On some approach legs? 3. What percentage of your signalized left-turn movements are... a. Protected-permissive? b. Permissive-only? 4. For the preceding left-turn modes, what percentage of the crossed pedestrian movements (see drawing on page 1) are controlled by pedestrian signal heads? a. Protected-permissive b. Permissive-only 5. Are you using the flashing yellow arrow display for any of your protected-permissive left turns? If so, what percentage of the protected-permissive movements? 6. What role does pedestrian volume play in the selection of... a. Left-turn mode? b. Left-turn phase sequence? 7. What percentage of your urban signalized intersections do you consider to have high pedestrian volumes? 8. How many of the high-pedestrian-volume intersections are in areas that you would describe as... a. Downtown / business district? b. School (K-8)? e. Other? c. High school? d. University? 9. Which of these sites are the most problematic with respect to pedestrians and left-turn movements? 10. At what percentage of your urban signalized intersections are you using... a. Pedestrian countdown signal heads? b. Leading pedestrian intervals? c. Other special pedestrian treatments? Issues Affecting Pedestrian Safety 11. How many pedestrian crashes per year do you have at your urban signalized intersections? 12. What percentage of the crashes are related to left-turn movements? 13. Consider the following pedestrian and signalized intersection characteristics. a. Left-turn volume d. Transit stop presence at intersection g. Coordination b. Approach speed e. Left-turn mode c. Pedestrian volume f. Phase sequence How influential are these site characteristics on pedestrian safety? Influence: 2 = very important, 1 = somewhat important, 0 = little or no importance How often are the characteristics considered in the design or operation of your signalized intersections? Discussion frequency: 2 = always discussed, 1 = sometimes discussed, 0 = rarely or never discussed 14. Consider the following pedestrian safety treatments. a. Add protected mode e. Prohibit left turns i. Provide leading ped. interval b. Add protected-permitted (PPLT) f. Add far-left signal head j. Provide ped. clear in green c. Add permissive omit with PPLT g. Provide ped. signal heads k. Add special signs or markings d. Change from leading to lagging h. Provide countdown heads l. Prohibit pedestrian crossing At what percentage of your signalized intersections have you used the treatments to improve ped. safety? Usage: Approximate percentage of intersections with the treatment How effective have they been in improving pedestrian safety? Effectiveness: 2 = very effective, 1 = somewhat effective, 0 = not effective at all 15. Do you use any criteria to decide where the previously discussed treatments should be installed? Figure 3-1. Interview Questions for Texas Agencies. 3-2

41 INTERVIEW QUESTIONS Inventory and Practice 1. What cities or states are you most familiar with regarding pedestrian practices at signalized intersections? 2. How many urban signalized intersections do you operate in your jurisdiction? 3. What role does pedestrian volume play in the selection of... a. Left-turn mode? b. Left-turn phase sequence? 4. What percentage of your urban signalized intersections do you consider to have high pedestrian volumes? 5. How many of the high-pedestrian-volume intersections are in areas that you would describe as... a. Downtown / business district? b. School (K-8)? e. Other? c. High school? d. University? 6. Which of these sites are the most problematic with respect to pedestrians and left-turn movements? Issues Affecting Pedestrian Safety 7. Consider the following pedestrian safety treatments. a. Add protected mode e. Prohibit left turns i. Provide leading ped. interval b. Add protected-permitted (PPLT) f. Add far-left signal head j. Provide ped. clear in green c. Add permissive omit with PPLT g. Provide ped. signal heads k. Add special signs or markings d. Change from leading to lagging h. Provide countdown heads l. Prohibit pedestrian crossing At what percentage of your signalized intersections have you used the treatments to improve ped. safety? Usage: Approximate percentage of intersections with the treatment How effective have they been in improving pedestrian safety? Effectiveness: 2 = very effective, 1 = somewhat effective, 0 = not effective at all 8. Do you use any criteria to decide where the previously discussed treatments should be installed? Figure 3-2. Interview Questions for Agencies Outside of Texas. Most of the interviewees were engineers, but a few engineering assistants and signal technicians were also interviewed. A total of 27 practitioners in Texas were visited in person, and an additional six in other states were contacted by telephone. The in-person meetings were about an hour in length, and the telephone calls were about a half-hour in length. The agencies represented by the interviewees are listed in Table

42 Table 3-1. Public Agencies Represented in the Practitioner Interview. Agency Representation Agency Name Number of Signalized Intersections Texas city Bryan 64 Tyler 145 Waco 174 Corpus Christi 250 Fort Worth 710 Austin 900 Dallas 1300 San Antonio 1230 TxDOT district Bryan 52 Corpus Christi 160 Tyler 188 Waco 220 Austin 300 San Antonio 500 Fort Worth 508 Dallas 339 Agency outside of Texas City of San Luis Obispo, California 57 City of Cambridge, Massachusetts 134 Salt Lake City Transportation Division, Utah 100 City of Madison, Wisconsin 249 City of Phoenix, Arizona 1082 City of Los Angeles, California 4500 Inventory and Practice To establish some overall perspective on the issues, the interviewees were asked questions relating to their inventory and general practice. These questions related to the number of urban signalized intersections within the interviewees jurisdiction, the use of crosswalks and pedestrian control, the selection of left-turn signalization parameters, and the location of intersections with high pedestrian volumes. The number of signalized intersections within each jurisdiction is listed in the last column of Table 3-1. Left-Turn Phasing The Texas interviewees were asked what percentage of their signalized left-turn movements are controlled with the protected-permissive or permissive-only modes. These left-turn movements are of the greatest interest because conflicts between pedestrians and left-turning vehicles can occur during the permissive period. The distribution of left-turn modes used by the agencies is provided 3-4

43 in the following list. As indicated in this list, 75 percent of the left-turn movements are operated with modes that result in possible conflict between left-turning vehicles and pedestrians.! Protected-permissive mode 51 percent.! Protected, split phasing, or other 25 percent.! Permissive 24 percent. All of the Texas interviewees indicated that pedestrian volume is rarely considered in the selection of left-turn mode. The selection of left-turn mode is typically based on vehicular volume and intersection geometry. More generally, they indicated that pedestrian safety is considered in two situations. One situation is where pedestrian crash data indicate the existence of a safety problem that can be addressed by protected left-turn operation. A second situation applies to intersections near schools. Several interviewees stated that they consider providing left-turn protection at school intersections or intersections near special-event facilities, even if vehicular volumes and intersection geometry do not justify providing left-turn protection. Two city engineers use the permissive period omit with protected-permissive mode at school intersections during school hours. All of the Texas interviewees stated that pedestrian volume plays little or no role in the selection of left-turn phase sequence (i.e., leading versus lagging). Phase sequence is usually chosen either by default (i.e., the agency always uses a certain phase sequence), or based on coordination issues (i.e., to maximize progression bandwidth). One interviewee stated that he had tried changing a left-turn phase from leading to lagging to improve pedestrian safety, but changed it back to leading because he did not believe it yielded any benefit. Another interviewee stated that he changed a left-turn phase from leading to lagging at one intersection near a university in response to complaints from the public, and that the complaints stopped after he implemented the change. Two of the interviewees from other states indicated that they consider pedestrian volume in the selection of left-turn control. In one city, the protected-permissive mode is used if the left-turn cycle failure rate is more than 80 percent and the delay to the opposing through movement is 40 s/veh or less. With this guidance, pedestrian volume can indirectly influence the choice of left-turn control to the extent that pedestrian timing affects the left-turn cycle failure rate and the opposing through movement delay. In another city, pedestrian volumes are collected and used in a simulation model to quantify performance measures. This city also conducts field visits to assess whether gaps in the pedestrian stream are sufficiently large to serve left-turning vehicles. They may consider the protected-permissive mode if adequate gaps are too infrequent. This assessment is done on a qualitative basis. Most of the Texas interviewees responded that their agencies are not using the flashing yellow arrow signal display for left-turn movements. Of all the contacted agencies, only the Cities of Tyler and San Antonio are using the flashing yellow arrow display. Tyler uses the flashing yellow arrow display for 40 percent of their protected-permissive left-turn movements, while San Antonio uses this display for 2 percent of their protected-permissive movements. 3-5

44 Pedestrian Control The Texas interviewees were asked about the provision of crosswalks at their urban signalized intersections. Specifically, they were asked to indicate what percentage of their intersections had crosswalks on all, some, or none of their approach legs. The Texas interviewees responses to the questions about crosswalks are summarized in Table 3-2. Their responses indicate that crosswalks are provided more frequently by cities than by TxDOT districts. Cities provide crosswalks on some or all approach legs at 84 percent of their intersections, while TxDOT districts do so at about 64 percent of their intersections. Crosswalks are provided more often by cities because more of the city-operated signals are in areas with higher population density and pedestrian demand. Table 3-2. Crosswalk Provision Trends at Urban Signalized Intersections. Intersection Crosswalks Percentage of Intersections by Texas Agency City TxDOT Overall Crosswalks on all legs Crosswalks on some legs No crosswalks Total: The Texas interviewees indicated that their agencies will occasionally omit crosswalks because of lack of pedestrian demand, a desire to avoid conflict with vehicles, or the existence of site-specific constraints that prevent the provision of crosswalks. For example, at three-leg intersections, it is common to provide only one crosswalk across the major street. This approach eliminates the conflict between pedestrians and vehicles turning left from the minor street. Crosswalks are also usually omitted in the inside of diamond interchanges. All agencies provide crosswalks that are compliant with Americans with Disabilities Act (ADA) requirements. They sometimes omit crosswalks if pedestrian signal heads are not present. The Texas interviewees were asked how often they provide pedestrian signal heads for the crossed pedestrian movements for the various left-turn modes. The crossed pedestrian movement is the movement that potentially conflicts with left-turning vehicles during the green signal indication. The provision of pedestrian signal heads for these pedestrian movements indicates recognition of the potential conflict and a desire to provide basic pedestrian control. The responses indicate that about 74 percent of pedestrian movements that are crossed by protected-permissive left-turn movement are provided with pedestrian signal heads. About 62 percent of pedestrian movements that are crossed by permissive-only left-turn movements are provided with signal heads. Description of Intersections with High Pedestrian Volumes The interviewees indicated that 10 to 15 percent of their urban signalized intersections have high pedestrian volumes. The percentage was slightly higher for cities (15 percent) than for TxDOT 3-6

45 districts (12 percent). The interviewees were asked to classify their intersections by providing the approximate number of intersections that are located around downtown areas or business districts, K-8 schools, high schools, universities, or other area types. The total count of intersections within these area type categories is shown in Table 3-3. For all area type categories, more high-pedestrian-volume signalized intersections are operated by cities than by TxDOT districts. Table 3-3. Classification of High-Pedestrian-Volume Intersections. Location of High- Percentage of Intersections by Agency Pedestrian-Volume Intersections City TxDOT Overall Downtown, business area K-8 school High school University Other Total: The bulk of the high-pedestrian-volume intersections are located within downtown areas or other business districts. A significant number also exist near K-8 schools. A total of about 150 intersections were located in area types that were described as other. These intersections included tourist areas, facilities that host special events, hike and bike trail crossings that occur at signalized intersections, hospitals, and assisted-living facilities. The interviewees were asked which intersections are the most problematic in terms of conflicts between pedestrians and left-turning vehicles. Their responses varied widely, though they gave different reasons for each area type to be problematic. The area type categories are discussed in greater detail in the following paragraphs. Business District. Six interviewees considered their downtown or business district intersections to be the most problematic for pedestrians and left-turning vehicles. They explained that left-turn movements in downtown areas are often operated in permissive-only mode. At times when pedestrian volumes are high, pedestrians tend to dominate the crosswalks, making it difficult for drivers to find safe gaps to execute left turns. This condition may cause drivers to attempt left turns using pedestrian gaps that are too narrow. Other issues include the presence of many distractions for drivers in downtown areas and the provision of light rail transit service. School. Eleven interviewees considered intersections near schools to be the most problematic. Signalized intersections near schools tend to have safety and operational problems because they experience high volumes of both pedestrians and vehicles simultaneously. Another common problem is that pedestrians at school intersections often do not comply with the pedestrian control (i.e., they do not wait for the WALK indication) or they place too much trust in the pedestrian control (i.e., they cross as soon as they are given a WALK indication, without first 3-7

46 checking for vehicles). These mistakes tend to be most prevalent among younger pedestrians, like students at K-8 schools. Issues Affecting Pedestrian Safety The Texas interviewees were asked to provide input on which intersection-related factors they think have the greatest influence on pedestrian safety at signalized intersections. Their responses are summarized in the following subsections. Characteristics Affecting Pedestrian Safety The Texas interviewees were asked whether various traffic and signalization characteristics had an influence on pedestrian safety. Some of the characteristics describe the intersection control, while others describe the users (vehicle traffic or pedestrians). For each characteristic, interviewees indicated (1) the amount of influence it has on safety and (2) the frequency it is considered as part of intersection design or operation. A numeric score was used to indicate the interviewee s response. Pedestrian volume ranks highest in terms of both influence and discussion frequency. The Texas interviewees believe that approach speed, left-turn volume, and left-turn mode are relatively influential on pedestrian safety. In contrast, left-turn phase sequence and signal coordination were indicated to have little influence and are rarely discussed. Use of Pedestrian Safety Treatments The interviewees were asked if they had used various safety treatments. Most of the treatments considered are listed in the first column of Table 2-3 in Chapter 2. The following additional treatments were added to this list:! Use permissive period omit with protected-permissive mode.! Prohibit left turns.! Add far-left supplemental signal head.! Provision of conventional pedestrian heads.! Provision of countdown pedestrian signal heads. The interviewees were asked to indicate the percentage of their urban signalized intersections at which they have used these treatments specifically to improve pedestrian safety. Instances where the treatment was used for reasons other than pedestrian safety were not considered. For those treatments that they have used, the interviewees were asked how they would rate the effectiveness of the treatments in improving pedestrian safety. A numeric score was used to indicate the interviewee s response. The two most commonly used treatments in Texas are (1) providing conventional pedestrian signal heads and (2) providing the pedestrian clear interval entirely during green. These two treatments are rated between somewhat effective and very effective in improving pedestrian safety. 3-8

47 Cities outside of Texas commonly use four treatments. Two of the four treatments are those identified in the previous paragraph. The other two commonly used treatments are (1) add far-left supplemental signal head and (2) provide countdown pedestrian signal head. The interviewees indicated that countdown heads are being used at more intersections each year. Another common treatment is prohibiting a pedestrian crossing. According to the interviewees, crossings are often prohibited on one leg of three-leg intersections to avoid conflict with left-turning vehicles from the minor street, on the inside of diamond interchanges, or at sites that do not comply with ADA requirements. This treatment is rated between somewhat effective and very effective in improving pedestrian safety. Criteria for Using Safety Treatments The interviewees were asked whether they use criteria or rules-of-thumb to decide where to use pedestrian safety treatments. The most commonly mentioned consideration regarding the decision to use a pedestrian safety treatment is citizen feedback and observations made during field visits. Criteria commonly cited include some consideration of pedestrian volume or evidence of pedestrian demand (e.g., worn footpaths observed on the roadside), visibility and sight distance, ADA requirements, and presence of school facilities. Some agencies provide treatments near schools when requested by the school district. Some of the treatments are used simply as a matter of policy or current practice. For example, some cities or districts provide pedestrian signal heads at all new traffic signals. Summary Interviews with 27 practitioners were conducted regarding their agency s left-turn signalization and pedestrian control practices. According to information provided by the interviewees, about 75 percent of urban signalized left-turn movements are operated in protected-permissive or permissive-only modes, which result in potential conflicts between left-turning vehicles and pedestrians. On average, the interviewees indicated that pedestrian volume is the most influential intersection characteristic on pedestrian safety. Most of the state s signals with high pedestrian volume are located in cities, specifically in urban areas or near schools. Other characteristics that influence pedestrian safety include approach speed, left-turn volume, and left-turn mode. When asked about treatments used to improve pedestrian safety, the most frequently mentioned options included adding pedestrian signal heads (conventional or countdown), provision of the pedestrian clear interval entirely during green, and prohibiting crossings. The interviewees explained that they do have guidance to select left-turn mode, but the guidance does not incorporate pedestrian considerations. The interviewees do not have comprehensive guidance on when to implement special pedestrian safety treatments, and often implement these treatments in response to citizen feedback. 3-9

48 EXPERT PANEL DISCUSSION This part of the chapter documents an electronic discussion (e-discussion) that was conducted to assess pedestrian safety issues at urban signalized intersections. The e-discussion was conducted with an expert panel comprised of 12 state and city traffic engineering practitioners. The panel was provided with information about three signalized intersections where conflicts occurred between pedestrians and drivers executing permissive left-turn maneuvers. The purpose of the e-discussion was to obtain insight into the causes of the demonstrated conflicts, to assess their severity, and to identify treatments to reduce conflict frequency. Background The e-discussion was conducted using a survey website to minimize time demands on the participants. This approach was rationalized to result in greater participation in the expert panel discussion. It was also a convenient method for presenting visual materials to the panel members. During the interviews that were in a previous part of this chapter, interviewees were asked if they would be willing to participate in an expert panel discussion. Those who indicated willingness to participate were invited to join the e-discussion. There were a total of 21 invitations sent out, and 12 people participated. The agencies represented by the participants are identified in Table 3-4. Table 3-4. Public Agencies Represented in the Expert Panel Discussion. Agency Representation Agency Name Number of Signalized Intersections Texas city Austin 900 Waco 174 San Antonio 1230 TxDOT district Bryan 52 Corpus Christi 160 Waco 220 Austin 300 San Antonio 500 Fort Worth 508 Agency outside of Texas City of San Luis Obispo, California 57 City of Cambridge, Massachusetts 134 City of Los Angeles, California 4500 Website Content The website that was used to host the e-discussion had seven pages. The first page included background information about the purpose and length of the e-discussion. On the second page, the participants were given a list of treatments that may be used to reduce conflicts between pedestrians 3-10

49 and left-turning vehicles. On the next three pages, the participants were shown three different scenarios where conflicts occurred at signalized intersections. Each scenario consisted of site description information, video clips of a conflict event observed from two angles, and questions about the contributing factors to the demonstrated conflict and possible treatments. On the last two pages, the participants were asked to indicate which agency they represented, and then given a brief concluding statement thanking them for their participation. On page 2 of the e-discussion website, a list of treatments was provided, along with a graphic image to illustrate each treatment. These images included signal heads that are used for the different left-turn modes, examples of signs, or drawings indicating the type of phasing used. The participants were given the list of treatments identified in Table 2-3 of Chapter 2. Conflict event scenarios were shown on pages 3, 4, and 5 of the e-discussion. For each of the three scenarios, a conflict event was demonstrated using video clips. Each event was shown from the two viewing angles. One view was provided by a tripod-mounted camcorder located on the curb end of the subject crosswalk and pointed toward the crosswalk, along a line parallel to the crosswalk. The second view was provided by a pole-mounted camcorder located on the curb diagonally opposite from the subject crosswalk. It provided a side view of pedestrians using the subject crosswalk. The expert panel members were also provided with plan-view drawings, street-level photographs, and information about the traffic volumes and signalization characteristics of the sites. This information was intended to help the expert panel members understand the nature of the problems so they could assess the contributing factors and identify treatments to alleviate the problems. The expert panel members were asked the questions listed in Figure 3-3 after they reviewed the site description information and the video clips of the conflict event. The questions were repeated for each of the three scenarios. Scenarios The scenarios are described in greater detail in the following three sections. The expert panel members responses to the questions about contributing factors and treatments are also provided. The members responses to the questions about conflict occurrence frequency is discussed in an aggregated manner in the next section because of the high degree of variability in the responses. Scenario 1 The intersection is located near a university campus, and experiences moderate pedestrian volumes and light left-turn volumes. The opposing intersection approach is a campus exit-only driveway that experiences light traffic volumes. The subject left-turn movement is operated in the permissive mode. Conventional pedestrian signal heads are provided, but pedestrian push buttons are not provided. 3-11

50 E-DISCUSSION QUESTIONS Contributing Factors 1. Assume that events similar to the one shown in the video clip occur often at this site. What are the contributing factors to the problem? Choose up to three. a. Inadequate sight distance b. High vehicular volume c. High pedestrian volume d. Inadequate opportunity for drivers to make a permissive left turn (protected left-turn phase not provided or not long enough) e. Inadequate time for pedestrians to complete the crossing (walk and/or pedestrian clear intervals are too short for individual pedestrians, or more time is needed due to high pedestrian volume) f. Left-turn phase sequence violates pedestrians expectations g. Aggressive driving h. Other (please specify open-ended) Treatments 2. You have been directed to implement one or more treatments to address this problem. Which of the following treatments would you use? Choose up to three. a. Change the left-turn mode to protected-permissive b. Change the left-turn mode to protected-permissive with the flashing yellow arrow display c. Change the left-turn mode to protected d. Make the protected left-turn phase lagging instead of leading e. Provide a leading pedestrian interval f. Provide an exclusive pedestrian phase g. Provide the pedestrian clear interval entirely during green (i.e., do not use the yellow change and red clearance intervals as part of the pedestrian clearance time) h. Add turning vehicle warning signs and markings i. Remove the crosswalk and prohibit crossing at this location j. Other (please specify open-ended) Conflict Occurrence Frequency 3. How often would events like the one in the video clip need to occur before you would deem it necessary to treat the site? Consider only daytime cycles when pedestrians are present. a. Occurs in 1% of cycles b. Occurs in 5% of cycles c. Occurs in 25% of cycles d. Occurs in $ 50% of cycles e. This problem is never important enough to require treatment Figure 3-3. E-Discussion Questions. The video clip shows that, as the pedestrians started to cross, a left-turning driver pulls into the intersection to wait for a safe gap in the pedestrian stream. The driver starts steering into the turn at a time that there are no opposing through vehicles present, but is delayed because a group of pedestrians start crossing after the start of the flashing DON T WALK indication. While the left-turning driver moves very slowly through the vehicle-vehicle conflict area (see Figure 1-1), a slow-moving opposing through vehicle arrives and is forced to maneuver around the waiting left-turning vehicle. 3-12

51 Scenario 2 The intersection is located near a university campus. Both the pedestrian volume and the left-turn volume are high. The opposing intersection approach is a campus street that experiences light traffic volume. The subject left-turn movement is operated in protected-permissive mode with a leading sequence. Conventional pedestrian signal heads are provided, but pedestrian push buttons are not provided. The video clip shows that a large number of pedestrians are present on both corners. Many of them start crossing prior to the start of the WALK indication, especially those originating from the near-side corner. Some pedestrians also start crossing after the start of the flashing DON T WALK indication. As the pedestrian platoons from the near-side and far-side corners approach each other, several left-turning vehicles pass through the closing gap in pedestrians. The gap for the last vehicle is short enough that several approaching pedestrians slow while the vehicle passes. Scenario 3 The intersection is located in a downtown grid area, with several restaurants, retail stores, and parking lots nearby. Both the pedestrian volume and the left-turn volume are low. The subject left-turn movement is operated in permissive mode. Conventional pedestrian signal heads are provided, but pedestrian push buttons are not provided. The video clip shows that, as a single pedestrian is crossing, a fast-moving left-turning vehicle arrives. The left-turning driver apparently does not notice the pedestrian and has to stop abruptly to avoid hitting the pedestrian. The pedestrian stops walking when he sees the left-turning driver approaching. Findings The responses of the expert panel members were aggregated across the three scenarios presented in the e-discussion. Common themes emerging from the aggregated responses are summarized in the following paragraphs, along with their implications for guideline development. Contributing Factors The expert panel members aggregated responses regarding contributing factors are shown in Figure 3-4. Overall, the most commonly identified contributing factors for conflicts between pedestrians and left-turning vehicles are: aggressive driving, high pedestrian volume, and inadequate opportunity to turn. Turning opportunities decrease with increasing volumes of pedestrians or opposing through or right-turning vehicles. As turning opportunities become less frequent, aggressive driving behaviors on the part of left-turning drivers may increase. Hence, treatment selection guidelines that consider pedestrian and vehicular volumes are likely to account for the influence of the most commonly identified contributing factors. 3-13

52 Number of Responses Inadequate sight distance High vehicular volume Contributing Factor High pedestrian volume Inadequate opportunity to turn Inadequate time to cross Phase sequence confuses pedestrians Aggressive driving Aggressive pedestrians Other Figure 3-4. Factors Contributing to Conflicts. One expert panel member stated that turning vehicle speed increases with intersection size, and opined that permissively turning drivers are less likely to stop for pedestrians if their speed is higher. Guidelines for selecting left-turn mode often account for the speed of opposing through vehicles. However, these guidelines do not directly account for turning vehicle speed. It is possible that these two speeds are correlated, as both are likely influenced by intersection geometry and area type. Selected Treatments The expert panel members aggregated responses regarding treatments are shown in Figure 3-5. The four most commonly identified treatments include: provide a protected left-turn phase, provide a leading pedestrian interval, provide an exclusive pedestrian phase, and add turning vehicle signs and markings. These treatments are discussed in Chapter 2. One expert panel member observed that the subject left-turn movements depicted in the scenarios lacked turn bays. This member suggested restriping the intersection approach at one of the scenario intersections to provide a turn bay. He opined that left-turning drivers in shared left/through lanes are more likely to turn hastily because they think they are more likely to be rear-ended. Left-turning drivers in shared left-and-through lanes may also feel greater pressure to turn if they see a long queue forming behind them. Hence, it may be desirable to add provide exclusive left-turn bay to the list of treatments for alleviating conflicts between pedestrians and left-turning vehicles. This treatment would apply for cases when the permissive mode is retained. 3-14

53 Number of Responses Protected-permissive mode Protected left-turn mode Protected left-turn mode Provide a leading left-turn phase Treatment Make left-turn phase lagging Leading pedestrian interval Exclusive pedestrian phase Pedestrian clear within green Signs and markings Other Protected-permissive mode with permissive-period omit Figure 3-5. Rank of Selected Treatments. Another expert panel member opined that a permissive left-turn period should never be allowed to run concurrently with pedestrian service if the street being crossed is six or more lanes in width. This member expressed support for adding curb extensions and reducing lane counts on the street being crossed to reduce crossing distances. He also generally favored eliminating conflicts between pedestrians and left-turning vehicles by providing either protected left-turn mode or exclusive pedestrian phases. Two treatments that were not commonly supported by the expert panel members include providing protected-permissive mode and providing the pedestrian clear interval entirely during green. The expert panel members apparently did not see these treatments as being effective in reducing conflicts between pedestrians and left-turning vehicles. Conflict Occurrence Frequency For each of the three scenarios, the expert panel members were asked to indicate how often the demonstrated conflict would need to occur before they would consider the problem significant enough to justify treatment. They were asked to consider only daytime cycles with pedestrians present. In other words, cycles occurring during periods of negligible pedestrian demand (e.g., nighttime cycles) were not considered. The choices allowed are listed at the bottom of Figure 3-3. The expert panel members responses varied significantly for each specific scenario. However, when the responses were aggregated across the three scenarios, a useful threshold emerged. To compute this threshold, the five discrete choices listed at the bottom of Figure 3-3 were regarded as ranked intervals indicating the probability of conflict occurrence. The width of an interval was based on splitting the difference between the percentages of each adjacent choice. The expert panel members responses were counted for each choice and scenario. Then, the average 3-15

54 threshold conflict frequency was computed for each scenario using Equation 2. These thresholds represent the frequency at which the scenario s conflict would need to occur (in terms of percent of signal cycles) before the average expert panel member would deem it necessary to treat the site. where, T j = threshold conflict frequency for scenario j ( j = 1, 2, 3), percent of cycles. c i, j = response count for scenario j and interval i (i = 1, 2,..., 5). w = probability interval width, percent. Equation 2 was used to compute the averages shown in the last row of Table 3-5. The threshold conflict frequencies are 25, 20, and 25 percent of cycles, respectively, for the conflicts shown in scenarios 1, 2, and 3. The consistency of this percentage among scenarios suggests that treatments should generally be considered for a site if conflicts between pedestrians and left-turning vehicles can be observed in 25 percent or more of the signal cycles. (2) Table 3-5. Conflict Occurrence Response Analysis. Interval Description Interval Response Count by Scenario (c) (i) Width (w), % Scenario 1 Scenario 2 Scenario 3 1 Occurs in 1% of cycles Occurs in 5% of cycles Occurs in 25% of cycles Occurs in $ 50% of cycles Never important Average Threshold Conflict Frequency (T), %: Summary In addition to the exposure variables of pedestrian and vehicular volumes, the expert panel responses suggest that turning vehicle speed may be an appropriate consideration when selecting left-turn mode. Based on the expert panel responses shown in Figure 3-5, the provision of a protected-permissive left-turn mode and the provision of the pedestrian clear interval entirely during green do not appear to be highly regarded options to reduce conflicts between pedestrians and left-turning vehicles. The expert panel responses suggest that treatments should be considered if conflicts between pedestrians and left-turning vehicles occur in 25 percent or more of signal cycles. 3-16

55 CHAPTER 4. MODEL DEVELOPMENT AND DATA COLLECTION OVERVIEW This chapter describes the development of a data collection plan. The data elements to be collected were identified through the examination of models used to represent pedestrian and vehicle behavior. There are two components of the data collection plan. One component is focused on the plan for collecting vehicle operations data. The data collected will be used to quantify the effect of alternative pedestrian-safety-related treatments on vehicle operations. The second component is focused on the plan for collecting pedestrian safety data. The data collected will be used to quantify the effect of alternative pedestrian-safety-related treatments on pedestrian safety. This chapter consists of two parts. The first part describes the operations model and data collection plan. The second part describes the safety model and data collection plan. OPERATIONS MODEL AND DATA COLLECTION PLAN This part of the chapter describes the steps taken to develop a data collection plan to quantify the effect of alternative pedestrian-safety-related treatments on vehicle operations at signalized intersections. A simulation model is used to obtain the data. The first section summarizes the rationale for selecting a specific simulation model. The second section describes the elements of the data collection plan. Evaluation of Alternative Simulation Models This section summarizes the findings from an evaluation of two candidate simulation tools. The evaluation separately considered each tool s ability to quantify the effect of pedestrian treatments on vehicular performance. The features and capabilities discussed in this section are referenced to the specific versions of the tools evaluated (i.e., VISSIM 5.10 and Synchro 6.0). These findings may not be applicable to newer versions of these tools. Pedestrian Models There are two levels of pedestrian modeling in VISSIM. The first level is a standard pedestrian model which treats pedestrians as one type of vehicle. In the standard model, VISSIM allows users to create their own pedestrian vehicles. Users can modify the pedestrian vehicles characteristics to replicate pedestrian attributes. These characteristics include pedestrian volume, walking speed, critical gap, and pedestrian car-following logic. The pedestrian car-following logic allows pedestrians to follow each other with much less restriction than vehicular traffic. The second level is an advanced implementation of pedestrian modeling in VISSIM where the movement of pedestrians is based on the Social Force Model (SFM). Under SFM, pedestrians are able to move in all directions (including counter-flow situations), and the trips are based on designated origin-destination patterns. The main advantage of the SFM is the ability to model interactions among pedestrians. While the SFM module is much more realistic than the standard 4-1

56 pedestrian level model, it is not a standard add-on in the VISSIM version used. Based on this examination, the standard pedestrian model was rationalized to be appropriate for the evaluation of the pedestrian safety treatments identified in Chapter 2. Synchro does not explicitly treat pedestrians as separate objects. Pedestrian characteristics that are provided as input to the models include: pedestrian calls per hour and pedestrian crosswalk volume. The former input represents the number of pedestrian push button calls for the phase. The latter input represents the number of pedestrians per hour that conflict with the right-turn movement during permitted operation. Pedestrian-Vehicle Interactions VISSIM provides two alternative methods for modeling pedestrian-vehicle interactions. They are the priority-rules method and the conflict-area method. Both methods have unique advantages. With the priority-rules method, priority rules are used to define the right of way for non-signal-protected conflicting movements. Vehicles and pedestrians recognize the presence of each other, even if they are on different links or connectors. The conflict-area method is based on a defined conflict area at the intersection of two links. For each conflict area, the user can select which of the conflicting links has right of way (if any). As they approach a conflict area, drivers and pedestrians plan how to cross the conflict area. A yielding driver (or pedestrian) observes the approaching vehicles (or pedestrians) in the main stream and identifies an acceptable gap. Then, the driver (or pedestrian) plans an acceleration profile for the next few seconds that will allow him or her to cross the conflict area. Realistic representation of pedestrian-vehicle interactions is more critical to the evaluation of pedestrian treatments because it will influence the accuracy to which VISSIM can estimate the effect of pedestrians and various treatments on vehicle performance. The most appropriate method may depend on the treatment being evaluated. In Synchro, pedestrian-vehicle interactions are incorporated into the calculation of saturation flow rate for the conflicting vehicle traffic stream. It is based on the model used in the 2000 Highway Capacity Manual (1), which is discussed in Chapter 2. Modeling Pedestrian Treatments Table 4-1 summarizes the options for modeling pedestrian treatments with VISSIM and Synchro. Both simulation tools are shown to possess the features required for evaluating pedestrian treatments. 4-2

57 Table 4-1. Comparison of Simulation Tool Capabilities. Treatment VISSIM Synchro Treatments Based on Conversion from Permissive Mode Provide protected-permissive mode Yes Yes Provide protected-permissive mode using flashing Right-of-way rules must be Yes yellow signal display defined Provide protected left-turn mode Yes Yes Provide lagging phase sequence Yes Yes Treatments Used in Conjunction with Permissive Mode Provide leading pedestrian interval No, for ring-based control No Provide exclusive pedestrian phase Yes Yes Provide pedestrian clear interval entirely during green Yes Yes Add turning vehicle warning signs and markings not applicable not applicable Prohibit pedestrian crossing Yes Yes Measures of Effectiveness Both VISSIM and Synchro are capable of reporting operational performance measures such as vehicle delay, queue length, and stop rate. However, VISSIM features data collection points that can be strategically placed in the network to log event-based data. These event data can be post-processed to extract useful surrogate safety measures. This capability gives VISSIM a major advantage over Synchro because it can be used to evaluate both vehicle performance and pedestrian safety. Summary Both software tools were found to be capable of modeling pedestrian treatments that involve changing the left-turn operational mode and phase sequence. However, VISSIM uses more sophisticated pedestrian models, and it is slightly more capable at modeling some complex pedestrian safety treatments. Therefore, VISSIM was determined to be the more viable option relative to the objectives of the research. Data Collection Plan The operations data collection plan was designed to produce data that could be used to quantify the effect of alternative pedestrian-safety-related treatments on vehicle operations. Specifically, these data would be used to develop a quantitative method for predicting the change in intersection delay due to the implementation of a single or a combination of pedestrian treatments. Intersection delay is defined as the weighted average delay of all vehicular traffic movements at the intersection. The weight used in computing this average is the vehicular volume for the corresponding movement. Pedestrian delay was not considered in this analysis. This section describes the data collection plan based on the use of VISSIM. The plan consisted of three tasks. The first task was the development of test bed intersections for simulation 4-3

58 analysis. The second task was the development of modeling techniques for VISSIM so that each candidate pedestrian safety treatment could be accurately evaluated. The third task was the development of simulation run control criteria to guide the assembly of the simulation data. Each of these tasks is discussed in one of the following three subsections. Test Bed Intersections and Baseline Conditions The simulation modeling approach required the development of six test bed intersections, each with specified baseline conditions. Collectively, the test beds and baseline conditions are intended to represent typical isolated, fully actuated intersections in Texas. Table 4-2 and Table 4-3 summarize the baseline conditions. These conditions represent a range of typical intersection geometry, signal control, and traffic conditions. Table 4-2 indicates that six test bed intersections were developed to collectively represent a range of approach legs (i.e., 3 or 4) and approach through lanes (i.e., 2, 3, or 4). The test bed intersection is fully actuated with stop-line detection on all approaches. Table 4-2. Test Bed Geometric and Signalization Characteristics. Characteristics Geometry! Intersection legs: 3 and 4! Major street lanes: 2, 4, or 6 (total both directions)! Minor street lanes: 1(each direction)! Left-turn bay length: 250 ft or longer to prevent bay overflow! No exclusive right-turn lane. Signalization! Left-turn mode on major street: permissive! Left-turn mode on minor street: permissive! Phase sequence on major street: lead-lead! Phase sequence on minor street: lead-lead! Detection: 40-ft stop line loops, presence mode! Max. green for through phases: 50 s! Max. green for left-turn phases: 25 s! Min. green through phases: 10 s! Min. green for left-turn phases: 5 s! Yellow (Y): 3.0 s! Red clearance (Rc): 2.0 s! Passage time: 2.0 s! Min. recall: major through phases! Walk: 7.0 s! Ped. clear interval: crossing distance/3.5!(y+rc)! Ped. detectors and signal heads for all crosswalks! Right turn on red: allowed on all approaches! Leading pedestrian interval: no! Exclusive pedestrian phase: no Comments! Six different test beds were required to capture the proposed variation in intersection legs and lanes.! The baseline signalization characteristics were the same for all test beds.! The calculation for pedestrian clear is based on average pedestrian speed recommended in MUTCD 2009.! Pedestrians were modeled to the extent that they press the button and cross in the crosswalk at the intersection.! Left-turn phases were not used under baseline conditions where all left turns are permissive-only mode. 4-4

59 Characteristics Traffic Characteristics! Major approach volume (veh/h/ln): 400, 600, and 800! Minor approach volume: 30% and 50% of major approach volume! Turn percentages for four-leg intersection: " Left turn: 10% " Right turn: 10%! Turn percentages for three-leg intersection: " Left-turn from minor approach: 45% " Right-turn from minor approach: 55% " Left-turn from major approach: 10% " Right-turn from major approach: 10%! 2-way pedestrian crosswalk volume (p/h/crosswalk): 0, 100, and 400! Average pedestrian speed: 3.5 ft/s! Approach speed: 35 mi/h! Left- and right-turn speeds: 15 mi/h! Percent trucks: 0% Yielding Characteristics! Front and rear gaps for permissive lefts and opposing through: 0.5 s! Front and rear gaps for permissive lefts and pedestrians: 0.5 s! Front and rear gaps for permissive rights and pedestrians: 0.5 s Table 4-3. Test Bed Traffic Characteristics. Comments! The delay difference between non-zero and zero-pedestrian scenarios determines the amount of delay incurred by pedestrians.! The saturation flow rate is a result of a combination of parameters specified in the car-following models and cannot be explicitly defined in VISSIM.! Standard pedestrian feature in VISSIM does not allow bidirectional movements for the link. Therefore, the 2-way pedestrian volumes are coded as two separate directional links with the 2-way volume split in half for each direction.! Conflict zones in VISSIM were used to model vehicle and pedestrian interactions.! Front Gap: Minimum gap in seconds between the rear end of a vehicle/pedestrian on the priority movement and the front end of a vehicle on the yielding movement.! Rear Gap: Minimum gap in seconds between the rear end of a vehicle on the yielding movement and the front end of a vehicle/pedestrian on the priority movement. Table 4-3 indicates that there are three levels of major approach volume, two levels of minor approach volume, and three levels of pedestrian volume. These variations produce 18 unique levels of traffic and pedestrian volume. The full factorial design represents a total of 108 unique combinations (= 6 test beds 18 volume levels) for simulation analysis. The 108 combinations described in this section represent the base conditions. These conditions were established to provide a basis for estimating the effect of each pedestrian treatment identified in Table 2-3 of Chapter 2. Pedestrian Treatments Each pedestrian treatment listed in Table 2-3 of Chapter 2 was simulated in VISSIM by modifying the base conditions. Table 4-4 describes these modifications as well as the changes made to the test beds for each treatment. All of the pedestrian treatments were implemented using VISSIM s ring-barrier control mode, which is very similar to the dual-ring signal control logic used in U.S. controllers. 4-5

60 Table 4-4. Test Bed Modifications for Each Pedestrian Treatment. Treatment Modifications Changes to VISSIM Test Beds Treatments Based on Conversion from Permissive Mode 1. Provide protectedpermissive mode 2. Provide protectedpermissive mode using flashing yellow (FY) signal display 3. Provide protected leftturn mode 4. Provide lagging phase sequence! Left-turn mode on major street: protected-permissive! Left-turn mode on minor street: protected-permissive! Left-turn mode on major street: protected-permissive (FY operation)! Left-turn mode on minor street: protected-permissive (FY operation)! Left-turn mode on major street: protected-only! Left-turn mode on minor street: protected-only! Phase sequence on major street: lag-lag! Phase sequence on minor street: lag-lag! Left-turn mode on major street: lag-lag! Left-turn mode on minor street: lag-lag Treatments Used in Conjunction with Permissive Mode 5. Provide leading pedestrian interval (LPI) 6. Provide exclusive pedestrian phase 7. Provide pedestrian clear interval entirely during green 8. Add turning vehicle warning signs and markings 9. Prohibit pedestrian crossing! Leading pedestrian interval: 3 s for all crosswalks.! Add exclusive pedestrian phase.! Prohibit RTOR on all approaches.! Pedestrian clear on all crossings: crossing distance divided by 3.5 ft/s! All red: 0 s! Pedestrian clear on all crossings: crossing distance divided by 3.5 ft/s not applicable not applicable! Adding left-turn phases in the controller.! Map the left-turn detector to the corresponding left-turn phase.! Permissive (circular) green for left turns is driven by the concurrent through phase.! Same as treatment 1 but permissive (circular) green for left turns is driven by an overlap of opposing through and left.! Same as treatment 1 but no permissive green indication is provided.! Apply the changes as in treatment 1 and then modify the phase sequence in the controller from lead-lead to lag-lag.! The effect of this treatment will be a combination of treatments 1 and 4.! LPI is not directly implemented in the VISSIM ring-barrier controller.! To simulate LPI, the through phase is defined to have a delayed green overlap driven by its own phase. The standard through phases (2, 4, 6, 8) will drive both the pedestrian heads and the delayed green overlap. As a result, the green for each through phase will be delayed by the specified LPI.! Remove pedestrian phases from concurrent vehicle phases.! Remove RTOR on all approaches.! Add exclusive pedestrian phase after the barrier of the major street phases.! Invoke the scramble phase option in the controller.! Use walk interval of 7 s.! Update the pedestrian clear setting in the controller. not applicable not applicable 4-6

61 Simulation Approach Run Control. A total of 864 simulation runs were completed. This total represented 108 runs to define the performance of the base conditions and an additional 756 runs (= 7 treatments x 108 combinations) to define the performance of the treatments. One replication was used for each combination. Each simulation run consisted of a 5-minute warm-up period and then a 60-minute data collection period. The results from each set of 108 simulation runs for a treatment were paired with the corresponding 108 runs for the base conditions. A one-to-one pairing of the performance estimates for each combination was used to determine the effect of the treatment. This approach allowed the researchers to quantify the effect of a treatment on delay, in isolation of the effect of other changes or treatments. Intersection Delay. The delay measurement obtained from VISSIM is equivalent to control delay. It is computed as the difference between the actual travel time and the ideal travel time. The ideal travel time represents the travel time when there are no other vehicles present, and control devices have no influence on vehicle speed. Intersection delay is a volume-weighted average control delay for the intersection, where the delay to each intersection movement is weighted by its volume. In VISSIM, delay is based on one or more travel time sections. Delay is measured for all vehicles that pass through these travel time sections, independently of the vehicle classes selected in these travel time sections. To collect intersection delay in VISSIM, 12 travel time sections (three movements per leg) were placed in the test bed for the four-leg intersections while six sections (two movements per leg) were used in the case of three-leg intersections. Each travel time section logs the average vehicle travel time and the number of vehicles traversing the section. The delay for the segment was calculated based on individual or a group of travel time sections. A post-processing tool was developed to help with extracting and reducing the simulated travel time data. This tool was designed to extract travel time and other key data elements from the simulation output files. It was also designed to perform any necessary calculations. The results from the simulation runs were reported in a spreadsheet with the output for each run placed in one row of the spreadsheet. Issues That Prevented Use of Some Combinations. VISSIM does not directly model the leading pedestrian interval (LPI) operation. In modern controllers, pedestrian signal heads are actuated, and LPI is activated only when there are pedestrian calls. However, in VISSIM ring-barrier control, LPI is active every phase, regardless of whether a pedestrian is present. This limitation was overcome by not simulating the cases of zero pedestrian volume with the LPI treatment. The results from a series of initial simulation runs indicated that the permissive-only mode used in several of the treatments was unable to accommodate heavy left-turn volumes, particularly in combinations that included 800 veh/h/ln and six lanes on the major street. This caused the queued 4-7

62 vehicles in the left-turn bay to grow indefinitely with the simulation time (i.e., an unsteady state). Such cases were identified and excluded from subsequent analyses. SAFETY MODEL AND DATA COLLECTION PLAN This part of the chapter describes the steps taken to develop a data collection plan to support the development of a pedestrian-vehicle conflict prediction model. This model would be applicable to conflicts at signalized intersections. It would be used to evaluate the safety benefits of alternative pedestrian treatments. The data are collected during field studies at several intersections. The first section describes the development of predictive models to guide the formulation of the data collection plan. The second section describes the elements of the data collection plan. Model Development Two models are described in this section. One model predicts the frequency of pedestrianvehicle conflicts based on the legal pedestrian volume. The second model predicts the legal pedestrian volume based on the total pedestrian volume. One variable that was included in the second model is pedestrian delay. Thus, this section also describes a model for predicting pedestrian delay. Pedestrian Delay Pedestrians crossing the street at a signalized intersection can experience delay due to the signal operation. If this delay is excessive, some pedestrians may choose to cross illegally (i.e., jaywalk). The average pedestrian delay is computed using Equation 3. This equation is obtained from Chapter 18 - Pedestrians of the Highway Capacity Manual (1). where, d p = pedestrian delay due to signal, s/p. C = cycle length, s. g ped = pedestrian service time (= Walk + 4.0), s. Walk = walk interval duration, s. Pedestrian-Vehicle Conflict Model The model for predicting conflicts between legally crossing pedestrians and left-turn vehicles is shown as Equation 4. A similar model form is used for conflicts between left-turn vehicles and pedestrians crossing illegally. where, N co,l = frequency of pedestrian-vehicle conflicts involving legal pedestrians, conflicts/h. v lt = conflicting left-turn volume, veh/h. (3) (4) 4-8

63 v ped,l = legal pedestrian volume in the crosswalk (walking in either direction), p/h. X i = independent variable i. b i = calibration coefficients, i = 0, 1, 2,... The first two terms in parentheses represent the left-turn volume term and the pedestrian volume term, respectively. Their product represents the combined exposure to pedestrian-vehicle conflict. The legal pedestrian volume represents the count of legal pedestrians divided by the count period duration. The third term in the equation represents the risk of conflict, given this exposure. The independent variables X i in the third term represent any variables that are found during model calibration to have some influence on the risk of conflict. Example variables include crosswalk length, vehicle travel time to conflict area, phase sequence, left-turn mode, cycle length, pedestrian delay, portion of the yellow interval used for pedestrian clearance, and queue clearance time for the opposing through movement. Legal Pedestrian Volume Model The preceding section described a model that is based on the legal pedestrian volume. A model for estimating this volume is described in Equation 5. (5) with where, v ped,l = legal pedestrian volume in the crosswalk (walking in either direction), p/h. v ped,t = pedestrian volume in the crosswalk (walking in either direction), p/h. p p = probability that the WALK indication is presented (= 1.0 if pretimed or on ped. recall). P b = probability of a pedestrian pressing the detector button. C = cycle length, s. c i = calibration coefficients, i = 0, 1, 2,... The first term in Equation 5 represents total pedestrian volume for the crosswalk during the analysis period. It is considered to be an exposure term for this model. The second term represents the probability that the WALK indication is presented. It is also considered to be an exposure term for this model. This probability is 1.0 if the signal operates under pretimed control or the phase is set for pedestrian recall. Equation 6 is used to compute this probability for pedestrian actuated operation. The probability of a single pedestrian pressing the detector button reflects the tendency of some pedestrians not to use the detector button before crossing a street. Research indicates that about 51 percent of crossing pedestrians will push the button to place a call for pedestrian service (2). (6) 4-9

64 The third term represents the probability that a pedestrian using the crosswalk is legal, as defined in Chapter 1, given that the phase was called. It is a logistic regression model, similar to that used by Hubbard et al. (3) to estimate the probability of conflict. The independent variables X i in the third term represent any variables that are found to have some influence on pedestrian compliance with the signal. Example variables include crosswalk length, median presence, average pedestrian age, pedestrian delay, conflicting vehicle flow rate, and type of pedestrian signal display (i.e., symbol or countdown). The variable for pedestrian delay was considered because it is rationalized that the probability of a legal crossing will decrease with an increase in delay due to the signal. Data Collection Plan This section describes the data collection plan. It consists of three subsections. The first subsection identifies the variables collected. The second subsection identifies the study sites. The third subsection describes the data collection procedures followed during each field study. Database Elements The data collected during this study were used to calibrate the models described in the previous section. Table 4-5 lists the types of data that were needed for model calibration. The characteristics and conflict measures listed were counted each cycle and summarized for each 15- min count period. The cycle length was measured each cycle and similarly summarized. The other variables listed were quantified during a site survey conducted prior to the start of the study period. 4-10

65 Table 4-5. Variables in Conflict Database. Category Variable Vehicle Count of left-turn vehicles crossing subject crosswalk characteristics Count of left-turn vehicles crossing subject crosswalk during opposing through green interval Count of through and right-turn vehicles opposing subject left-turn movement Count of vehicles crossing crosswalk during steady DON T WALK indication Pedestrian Count of pedestrians in subject crosswalk (walking in either direction) by age group 1 characteristics Count of pedestrians that enter crosswalk (or attempted to enter it) during the WALK indication by age group 1 Count of pedestrians that enter the crosswalk during the flashing or steady DON T WALK indication Intersection Length of subject crosswalk (curb face to curb face) geometry Width of subject crosswalk (center of marking to center of marking) Number of lanes serving subject left-turn movement Number of receiving lanes for subject left-turn movement Number of opposing through lanes Skew angle Distance from left-turn stop line to representative crosswalk conflict point Presence of a median and its width Signalization Cycle length Walk interval duration for subject crosswalk Pedestrian clear interval duration for subject crosswalk Left-turn mode (permissive or protected-permissive) Phase sequence (permissive, leading, lagging, split) Use of a leading or lagging pedestrian interval Allowance of right-turn-on-red Control type (pretimed, full-actuated, semi-actuated, coordinated-actuated) Traffic control Speed limit of traffic crossing subject crosswalk devices Presence of count-down pedestrian display Conflict Count of legal pedestrian-vehicle conflicts in conflict area of subject crosswalk measures Count of illegal pedestrian-vehicle conflicts in conflict area of subject crosswalk Note: 1 - Age groups: Child (less than 12 years), Teen (12-17), Young (17-25), Adult (25-60), Senior (60 or more). Study Sites At total of 20 study sites were chosen from among those available in the cities of College Station and Austin. A site is defined to be one crosswalk. Each site is associated with one leftturning movement. This is the movement that crosses the crosswalk as it exits the intersection. The College Station sites are located close to the Texas A&M University campus. The Austin sites were chosen from the downtown city grid and the area surrounding the University of Texas campus. The sites are identified in Table 4-6. A more detailed description of these sites is provided in Appendix A. Collectively, these sites represent a range of vehicular and pedestrian volumes. 4-11

66 Table 4-6. Study Site Description. City Intersecting Street by Direction Subject Left- Turn North-South East-West Movement Phase Sequence Left-Turn Volume, veh/h Pedestrian Volume, p/h Austin Congress Ave. 4 th Street Northbound Perm. only Eastbound Perm. only Southbound Perm. only Westbound Perm. only Congress Ave. 5 th Street Eastbound Perm. only Congress Ave. 6 th Street Northbound Lagging Westbound Perm. only Congress Ave. 7 th Street Southbound Lagging Guadalupe St. 21 st Street Eastbound Perm. only Westbound Perm. only Guadalupe St. 22 st Street Eastbound Perm. only Guadalupe St. 24 st Street Eastbound Leading Guadalupe St. Dean Keeton St. Southbound Leading Guadalupe St. 27 st Street Westbound Perm. only San Jacinto Blvd. Dean Keeton St. Northbound Perm. only Southbound Perm. only Medical Arts St./ Dean Keeton St. Northbound Perm. only Robert Dedman Dr. Southbound Perm. only College Station Spence Street University Drive Northbound Perm. only Southbound Perm. only Data Collection Procedure The data collection at each site consisted of a site survey followed by a video recording of vehicle and pedestrian activity. The recordings took place during the two-hour period that spans the peak hour of pedestrian demand at the site. Data were not collected during rain events. Pedestrian-vehicle conflicts were recorded by video camera during each field study. Two video cameras were used at each site to ensure full coverage of the intersection. Figure 4-1 illustrates a typical two-camera setup for evaluating the northbound left-turn movement and associated conflict area. One camera was located in line with the subject crosswalk. This camera monitored the pedestrian signal head, left-turn signal head, and crosswalk area. The second video camera was used to monitor the left-turn stop line, left-turn vehicle travel path, vehicle brake lights, and crosswalk. Together, the two cameras were able to provide three-dimensional coverage of pedestrian activity in the conflict area. 4-12

67 a. Plan View of Video Camera Locations. b. Camera 1 Location. c. Camera 2 Location. Figure 4-1. Video Camera Locations. REFERENCES 1. Highway Capacity Manual. Transportation Research Board, National Research Council, Washington, D.C., Zegeer, C., K. Opiela, and M. Cynecki. Pedestrian Signalization Alternatives. Report FHWA/RD-83/102. Federal Highway Administration, Washington, D.C.,